Skip to content

Hypergraph is more powerful. Transform unstructured text into structured knowledge with LLMs. Graphs, hypergraphs, and spatio-temporal extractions — with one command.

Remove unused templates for construction, cybersecurity, fashion, marketing, music, politics, real estate, recruitment, software, sports, supply chain, and travel from the hyperextract project.

yifanfeng974e86348d7b8a
diff --git a/README.md b/README.md
index 714be00..8dfb8c5 100644
--- a/README.md
+++ b/README.md
@@ -88,9 +88,9 @@ The `hyperextract/templates` directory provides a wealth of domain-specific defi
| :--- | :--- | :--- |
| **General** | `General`, `IText2KG` | Versatile extraction for most daily knowledge. |
| **Science** | `Physics`, `Chemistry`, `Biology` | Capture experiments, elements, and reactions. |
-| **Industry** | `Finance`, `Legal`, `Cybersecurity` | Extract compliance, risks, and entity webs. |
-| **Entertainment** | `Game`, `Movie`, `Music` | Perfect for story plots and character networks. |
-| ... | | 20+ more domains in the [templates folder](hyperextract/templates) |
+| **Industry** | `Finance`, `Legal` | Extract compliance, risks, and entity webs. |
+| **Entertainment** | `Game`, `Movie`, `Food` | Perfect for story plots, character networks, and recipes. |
+| ... | | 15+ domains in the [templates folder](hyperextract/templates) |
---
diff --git a/README_ZH.md b/README_ZH.md
index 12a29b3..cfc5c0d 100644
--- a/README_ZH.md
+++ b/README_ZH.md
@@ -88,9 +88,9 @@ print(answer.content)
| :--- | :--- | :--- |
| **通用** | `General`, `IText2KG` | 适用于大多数日常知识提取。 |
| **科学** | `Physics`, `Chemistry`, `Biology` | 捕捉科学实验、元素、反应过程。 |
-| **行业** | `Finance`, `Legal`, `Cybersecurity` | 提取合规、风险点、实体关联。 |
-| **娱乐** | `Game`, `Movie`, `Music` | 完美支持游戏剧情、角色关系网。 |
-| ... | | 更多 20+ 领域见 [templates 目录](hyperextract/templates) |
+| **行业** | `Finance`, `Legal` | 提取合规、风险点、实体关联。 |
+| **娱乐** | `Game`, `Movie`, `Food` | 完美支持游戏剧情、角色关系网、食谱制作。 |
+| ... | | 更多 15+ 领域见 [templates 目录](hyperextract/templates) |
---
diff --git a/hyperextract/templates/en/README.md b/hyperextract/templates/en/README.md
index c2aa374..b00231e 100644
--- a/hyperextract/templates/en/README.md
+++ b/hyperextract/templates/en/README.md
@@ -12,16 +12,10 @@ Welcome to the **Hyper-Extract Domain Templates** library. This directory contai
| Folder | Description | Potential Use Cases |
| :--- | :--- | :--- |
| **`agriculture`** | Farming, livestock, and crop management. | Crop rotation, irrigation systems. |
-| **`art`** | Fine arts, museum collections, and provenance. | Artwork catalogs, historical art movements. |
-| **`automotive`** | Vehicles, manufacturing, and maintenance. | Car specs, maintenance logs. |
| **`biology`** | Life sciences, genetics, and taxonomy. | Protein interactions, species classification. |
-| **`business`** | General corporate profiles and strategy. | Company bios, strategic goals. |
| **`chemistry`** | Molecular structures and reactions. | Chemical compounds, lab experiments. |
-| **`construction`** | Building, BIM, and project planning. | Construction milestones, blueprint entities. |
-| **`cybersecurity`** | Threat intelligence and network defense. | STIX models, attack vectors. |
| **`education`** | Knowledge points and curriculum. | Course prerequisites, learning paths. |
| **`environment`** | Climate, ecology, and sustainability. | Weather patterns, carbon footprint tracking. |
-| **`fashion`** | Garments, trends, and collections. | Design lookbooks, seasonal trends. |
| **`finance`** | Banking, stocks, and investments. | Earnings reports, stock analysis. |
| **`food`** | Culinary, recipes, and nutrition. | Menu planning, recipe breakdown. |
| **`game`** | Game lore, mechanics, and quests. | NPC relations, quest trees, item crafting. |
@@ -29,19 +23,10 @@ Welcome to the **Hyper-Extract Domain Templates** library. This directory contai
| **`history`** | Events, timelines, and figures. | War campaigns, historical biographies. |
| **`legal`** | Law, contracts, and compliance. | Contract obligations, court rulings. |
| **`literature`** | Fiction, poetry, and character arcs. | Character maps, literary themes. |
-| **`marketing`** | Campaigns, SEO, and personas. | Customer journeys, brand archetypes. |
| **`medicine`** | Western and clinical medicine. | Diagnosis, symptom mapping, pharma. |
| **`movie`** | Film, TV, and scripts. | Screenplay beats, cast & crew networks. |
-| **`music`** | Theory, albums, and discography. | Genre evolution, tracklists. |
| **`news`** | Media and trending topics. | Investigative boards, causality analysis. |
| **`physics`** | Particle physics, astronomy, and circuits. | Celestial systems, circuit paths. |
-| **`politics`** | Governance, policy, and geopolitics. | Legislative tracking, election maps. |
-| **`real_estate`** | Property, leasing, and zoning. | Property listings, floor plans. |
-| **`recruitment`** | HR, resumes, and hiring. | Skill mapping, interview assessment. |
-| **`software`** | Architecture, APIs, and DevOps. | Microservices, software dependencies. |
-| **`sports`** | Tournaments, matches, and stats. | Match event logs, player profiles. |
-| **`supply_chain`**| Logistics and inventory flow. | Supplier tiering, logistics routes. |
-| **`travel`** | Tourism and itineraries. | Trip planning, attraction reviews. |
## Usage
Simply import the template you need:
diff --git a/hyperextract/templates/en/__init__.py b/hyperextract/templates/en/__init__.py
index ada7ed1..db5ed8b 100644
--- a/hyperextract/templates/en/__init__.py
+++ b/hyperextract/templates/en/__init__.py
@@ -2,20 +2,10 @@ from .general.knowledge_graph import KnowledgeGraph
from .general.concept_map import ConceptMap
from .general.logic_graph import LogicGraph
from .agriculture.crop import CropStatusGraph
-from .agriculture.supply_chain import AgriSupplyChainGraph
from .agriculture.soil import SoilNutritionGraph
-from .art.history import ArtHistoryGraph
-from .art.style import ArtStyleGraph
-from .art.exhibition import ExhibitionGraph
-from .automotive.vehicle import VehicleSpecGraph
-from .automotive.diagnosis import VehicleDiagnosisGraph
-from .automotive.market import AutoMarketGraph
from .biology.taxonomy import TaxonomyGraph
from .biology.network import BiologicalNetwork
from .biology.metabolic import MetabolicHypergraph
-from .business.equity import EquityStructureGraph
-from .business.strategy import StrategicChainGraph
-from .business.ecosystem import MarketEcosystemHyper
from .chemistry.molecular import MolecularStructureGraph
from .chemistry.reaction import ChemicalReactionHyper
from .chemistry.protocol import LabProtocolTemporal
@@ -25,20 +15,10 @@ __all__ = [
"ConceptMap",
"LogicGraph",
"CropStatusGraph",
- "AgriSupplyChainGraph",
"SoilNutritionGraph",
- "ArtHistoryGraph",
- "ArtStyleGraph",
- "ExhibitionGraph",
- "VehicleSpecGraph",
- "VehicleDiagnosisGraph",
- "AutoMarketGraph",
"TaxonomyGraph",
"BiologicalNetwork",
"MetabolicHypergraph",
- "EquityStructureGraph",
- "StrategicChainGraph",
- "MarketEcosystemHyper",
"MolecularStructureGraph",
"ChemicalReactionHyper",
"LabProtocolTemporal",
diff --git a/hyperextract/templates/en/art/__init__.py b/hyperextract/templates/en/art/__init__.py
deleted file mode 100644
index 96cb725..0000000
--- a/hyperextract/templates/en/art/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .provenance import ProvenanceGraph
-from .movement import ArtMovementGraph
-from .exhibition import ExhibitionGraph
-
-__all__ = [
- "ProvenanceGraph",
- "ArtMovementGraph",
- "ExhibitionGraph",
-]
diff --git a/hyperextract/templates/en/art/exhibition.py b/hyperextract/templates/en/art/exhibition.py
deleted file mode 100644
index 1ac4a8c..0000000
--- a/hyperextract/templates/en/art/exhibition.py
+++ /dev/null
@@ -1,152 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class ExhibitionEntity(BaseModel):
- """An element of an art exhibition (Curator, Theme, Work, Venue, Sponsor)."""
-
- name: str = Field(description="Name of the exhibition, artist, or curator.")
- type: str = Field(
- description="Type: 'Exhibition', 'Theme', 'Artwork', 'Curator', 'Venue', 'Partner'."
- )
- details: Optional[str] = Field(
- description="Exhibition dates, conceptual notes, or location."
- )
-
-
-class ExhibitionRelation(BaseModel):
- """Connectivity within an exhibition's planning and execution."""
-
- source: str = Field(description="The organizing or controlling entity.")
- target: str = Field(description="The related work or participant.")
- relation_type: str = Field(
- description="Type: 'Curated by', 'Featured in', 'Hosted at', 'Sponsored by', 'Thematically linked to'."
- )
- context: Optional[str] = Field(
- description="Specific room, loan conditions, or thematic sub-group."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-EXHIBITION_GRAPH_PROMPT = (
- "You are a museum curator and exhibition designer. Map the structure of an art exhibition.\n\n"
- "Guidelines:\n"
- "- Identify the exhibition title, its central themes, and the participating artists/works.\n"
- "- Connect works to their thematic sections and the curators responsible.\n"
- "- Capture logistics like venues and sponsors."
-)
-
-EXHIBITION_NODE_PROMPT = "Extract exhibition components: titles, venues, themes, specific artworks, and key personnel (curators, designers)."
-
-EXHIBITION_EDGE_PROMPT = "Link items within the exhibition. Show which artworks belong to which theme and identify the hosting venue and sponsors."
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class ExhibitionGraph(AutoGraph[ExhibitionEntity, ExhibitionRelation]):
- """
- Template for curatorial planning and exhibition analysis.
-
- Tracks the relationships between artworks, themes, curators, and venues.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = ExhibitionGraph(llm_client=llm, embedder=embedder)
- >>> text = "The 'Surrealist Dreams' exhibition was curated by Alice Smith at the Tate Modern."
- >>> graph.feed_text(text)
- >>> print(graph.edges) # Exhibition connections
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the ExhibitionGraph template.
-
- Args:
- llm_client: The language model client for extraction.
- embedder: The embedding model for deduplication.
- extraction_mode: "one_stage" or "two_stage".
- chunk_size: Max characters per chunk.
- chunk_overlap: Overlap between chunks.
- max_workers: Parallel processing workers.
- verbose: Enable progress logging.
- **kwargs: Extra arguments for AutoGraph.
- """
- super().__init__(
- node_schema=ExhibitionEntity,
- edge_schema=ExhibitionRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.strip()}"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=EXHIBITION_GRAPH_PROMPT,
- prompt_for_node_extraction=EXHIBITION_NODE_PROMPT,
- prompt_for_edge_extraction=EXHIBITION_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: ExhibitionEntity) -> str:
- info = f" ({node.type})" if getattr(node, "type", None) else ""
- return f"{node.name}{info}"
-
- def edge_label_extractor(edge: ExhibitionRelation) -> str:
- return f"{edge.relation_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/art/movement.py b/hyperextract/templates/en/art/movement.py
deleted file mode 100644
index 1558901..0000000
--- a/hyperextract/templates/en/art/movement.py
+++ /dev/null
@@ -1,157 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class MovementEntity(BaseModel):
- """An art movement, style, artist, or influential work."""
-
- name: str = Field(
- description="Name of the movement (e.g., Impressionism) or Artist."
- )
- category: str = Field(
- description="Category: 'Movement', 'Style', 'Artist', 'Manifesto', 'Masterpiece'."
- )
- attributes: Optional[str] = Field(
- description="Key characteristics, materials, or timeframe."
- )
-
-
-class ArtInfluenceEdge(BaseModel):
- """Influence, derivation, or membership relations in art history."""
-
- source: str = Field(description="The precursor or movement.")
- target: str = Field(description="The follower or specific work/artist.")
- relation: str = Field(
- description="Relation: 'Influenced', 'Founder of', 'Member of', 'Reaction against', 'Derived from'."
- )
- details: Optional[str] = Field(
- description="Specific stylistic techniques or historical events."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-MOVEMENT_GRAPH_PROMPT = (
- "You are an art historian specializing in the evolution of artistic styles. Extract a graph of art movements and their influences.\n\n"
- "Guidelines:\n"
- "- Identify distinct art movements and the artists associated with them.\n"
- "- Map how one movement influenced another (e.g., Post-Impressionism reacting to Impressionism).\n"
- "- Capture the core philosophies and visual characteristics defined in manifestos or works."
-)
-
-MOVEMENT_NODE_PROMPT = "Extract art movements, styles, associated artists, and pivotal artworks. Describe their visual language and key theories."
-
-MOVEMENT_EDGE_PROMPT = (
- "Define connections between movements and individuals. Use relations like 'Reaction against' or 'Influenced'. "
- "Note specific techniques passed between generations."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class ArtMovementGraph(AutoGraph[MovementEntity, ArtInfluenceEdge]):
- """
- Template for mapping art movements, stylistic influences, and artist school affiliations.
-
- Useful for educational applications and art historical analysis.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = ArtMovementGraph(llm_client=llm, embedder=embedder)
- >>> text = "Impressionism, led by Monet, was later challenged by the structured approach of Post-Impressionism."
- >>> graph.feed_text(text)
- >>> print(graph.nodes) # Movement and Artist details
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the ArtMovementGraph template.
-
- Args:
- llm_client: The language model client for extraction.
- embedder: The embedding model for deduplication.
- extraction_mode: "one_stage" or "two_stage".
- chunk_size: Max characters per chunk.
- chunk_overlap: Overlap between chunks.
- max_workers: Parallel processing workers.
- verbose: Enable progress logging.
- **kwargs: Extra arguments for AutoGraph.
- """
- super().__init__(
- node_schema=MovementEntity,
- edge_schema=ArtInfluenceEdge,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}--({x.relation.lower()})-->{x.target.strip()}"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=MOVEMENT_GRAPH_PROMPT,
- prompt_for_node_extraction=MOVEMENT_NODE_PROMPT,
- prompt_for_edge_extraction=MOVEMENT_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: MovementEntity) -> str:
- info = f" ({node.category})" if getattr(node, "category", None) else ""
- return f"{node.name}{info}"
-
- def edge_label_extractor(edge: ArtInfluenceEdge) -> str:
- return f"{edge.relation}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/art/provenance.py b/hyperextract/templates/en/art/provenance.py
deleted file mode 100644
index 584c3fc..0000000
--- a/hyperextract/templates/en/art/provenance.py
+++ /dev/null
@@ -1,158 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class ArtEntity(BaseModel):
- """An entity in the art world (e.g., Artwork, Artist, Museum, Collector, Auction House)."""
-
- name: str = Field(description="Name of the artwork, person, or organization.")
- category: str = Field(
- description="Category: 'Artwork', 'Artist', 'Collector', 'Institution', 'Location'."
- )
- description: Optional[str] = Field(
- description="Medium, creation date, or historical status."
- )
-
-
-class ProvenanceEdge(BaseModel):
- """The transfer of ownership or location of an artwork."""
-
- source: str = Field(description="Previous owner or starting location.")
- target: str = Field(description="New owner or destination.")
- transfer_type: str = Field(
- description="Type: 'Acquired by', 'Sold to', 'Donated to', 'Lent to', 'Commissioned by'."
- )
- specification: Optional[str] = Field(
- description="Date of transfer, sale price, or auction details."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-PROVENANCE_GRAPH_PROMPT = (
- "You are a fine arts historian and provenance researcher. Extract an artwork's ownership history graph.\n\n"
- "Guidelines:\n"
- "- Identify the core artwork and all its historical owners (Collectors, Museums, Kings).\n"
- "- Map the chronological sequence of transfers (Sales, Inheritances, Looting, Donations).\n"
- "- Capture specific dates and monetary values if mentioned."
-)
-
-PROVENANCE_NODE_PROMPT = (
- "Extract art entities: identify the artwork itself, the artists, collectors, museums, and auction houses. "
- "Note the medium of the artwork and roles of the people/institutions."
-)
-
-PROVENANCE_EDGE_PROMPT = (
- "Establish the chain of title. Connect previous owners to current owners using transfer types like 'Sold to' or 'Donated to'. "
- "Include time periods and transaction details in the specification."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class ProvenanceGraph(AutoGraph[ArtEntity, ProvenanceEdge]):
- """
- Template for tracking the ownership history (provenance) of artworks and artifacts.
-
- Essential for art authentication, museum archives, and historical research.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI(model="gpt-4o")
- >>> embedder = OpenAIEmbeddings()
- >>> graph = ProvenanceGraph(llm_client=llm, embedder=embedder)
- >>> text = "The painting was acquired by the Louvre in 1815 from the Borghese collection."
- >>> graph.feed_text(text)
- >>> print(graph.edges) # Extracted ownership transfer
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the ProvenanceGraph template.
-
- Args:
- llm_client: The language model client for extraction.
- embedder: The embedding model for deduplication.
- extraction_mode: "one_stage" or "two_stage".
- chunk_size: Max characters per chunk.
- chunk_overlap: Overlap between chunks.
- max_workers: Parallel processing workers.
- verbose: Enable progress logging.
- **kwargs: Extra arguments for AutoGraph.
- """
- super().__init__(
- node_schema=ArtEntity,
- edge_schema=ProvenanceEdge,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}--({x.transfer_type.lower()})-->{x.target.strip()}"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=PROVENANCE_GRAPH_PROMPT,
- prompt_for_node_extraction=PROVENANCE_NODE_PROMPT,
- prompt_for_edge_extraction=PROVENANCE_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: ArtEntity) -> str:
- info = f" ({node.category})" if getattr(node, "category", None) else ""
- return f"{node.name}{info}"
-
- def edge_label_extractor(edge: ProvenanceEdge) -> str:
- return f"{edge.transfer_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/automotive/__init__.py b/hyperextract/templates/en/automotive/__init__.py
deleted file mode 100644
index 0184d1b..0000000
--- a/hyperextract/templates/en/automotive/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .car_spec import CarSpecGraph
-from .maintenance import CarMaintenanceGraph
-from .manufacturing import AutomotiveMfgGraph
-
-__all__ = [
- "CarSpecGraph",
- "CarMaintenanceGraph",
- "AutomotiveMfgGraph",
-]
diff --git a/hyperextract/templates/en/automotive/car_spec.py b/hyperextract/templates/en/automotive/car_spec.py
deleted file mode 100644
index 70a20c6..0000000
--- a/hyperextract/templates/en/automotive/car_spec.py
+++ /dev/null
@@ -1,141 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class CarComponent(BaseModel):
- """A vehicle component or specification category (Engine, Battery, ADAS, Transmission)."""
- name: str = Field(description="Name of the car model or specific component.")
- category: str = Field(
- description="Category: 'Model', 'Powertrain', 'Chassis', 'Safety System', 'Infotainment'."
- )
- specification: Optional[str] = Field(description="Detailed specs (e.g., Horsepower, Capacity, Version).")
-
-class CarSystemRelation(BaseModel):
- """The hierarchy or technical relationship between car systems."""
- source: str = Field(description="Parent system or vehicle model.")
- target: str = Field(description="Sub-component or feature.")
- relation_type: str = Field(
- description="Type: 'Equipped with', 'Powered by', 'Integrated with', 'Compatible with'."
- )
- performance_note: Optional[str] = Field(description="Benchmark data or integration details.")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CAR_SPEC_GRAPH_PROMPT = (
- "You are an automotive engineer and technical reviewer. Extract a technical specification graph for a vehicle.\n\n"
- "Guidelines:\n"
- "- Identify the vehicle model and its primary systems (Engine, Motor, Battery, ADAS).\n"
- "- Map the hierarchy of components (e.g., Battery contains Cells, Engine uses Turbocharger).\n"
- "- Capture quantifiable specs like 0-60 mph, Range, Peak Power, and Torque."
-)
-
-CAR_SPEC_NODE_PROMPT = (
- "Extract car models and technical components. Focus on names and their measurable specifications (e.g., 500hp, 80kWh)."
-)
-
-CAR_SPEC_EDGE_PROMPT = (
- "Define technical dependencies. Show which components belong to which car system. Use relations like 'Powered by' or 'Equipped with'."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class CarSpecGraph(AutoGraph[CarComponent, CarSystemRelation]):
- """
- Template for extracting vehicle technical data and specification hierarchies.
-
- Ideal for competitive analysis, e-commerce car data, and technical documentation.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = CarSpecGraph(llm_client=llm, embedder=embedder)
- >>> text = "The Tesla Model 3 Performance features a dual-motor AWD system delivering 510 hp."
- >>> graph.feed_text(text)
- >>> print(graph.nodes) # Vehicle and component specs
- """
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any
- ):
- """
- Initialize the CarSpecGraph template.
-
- Args:
- llm_client: The language model client for extraction.
- embedder: The embedding model for deduplication.
- extraction_mode: "one_stage" or "two_stage".
- chunk_size: Max characters per chunk.
- chunk_overlap: Overlap between chunks.
- max_workers: Parallel processing workers.
- verbose: Enable progress logging.
- **kwargs: Extra arguments for AutoGraph.
- """
- super().__init__(
- node_schema=CarComponent,
- edge_schema=CarSystemRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.strip()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CAR_SPEC_GRAPH_PROMPT,
- prompt_for_node_extraction=CAR_SPEC_NODE_PROMPT,
- prompt_for_edge_extraction=CAR_SPEC_EDGE_PROMPT,
- **kwargs
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: CarComponent) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.name }{info}"
-
- def edge_label_extractor(edge: CarSystemRelation) -> str:
- return f"{ edge.relation_type }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/automotive/maintenance.py b/hyperextract/templates/en/automotive/maintenance.py
deleted file mode 100644
index 603c387..0000000
--- a/hyperextract/templates/en/automotive/maintenance.py
+++ /dev/null
@@ -1,137 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class MaintenanceItem(BaseModel):
- """A maintenance task or vehicle part (Oil Change, Brake Pad, Timing Belt)."""
- item_name: str = Field(description="Name of the part or maintenance service.")
- category: str = Field(description="Category: 'Fluid', 'Wearable Part', 'Inspection', 'Software'.")
- interval: Optional[str] = Field(description="Suggested mileage or time interval (e.g., 5,000 miles).")
-
-class MaintenanceSequence(BaseModel):
- """The dependency or sequence of maintenance actions."""
- source: str = Field(description="Prerequisite component or initial state.")
- target: str = Field(description="Maintenance task or replacement part.")
- action: str = Field(description="Action: 'Replace', 'Inspect', 'Flush', 'Calibrate'.")
- precaution: Optional[str] = Field(description="Safety warnings or required tools.")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-MAINTENANCE_GRAPH_PROMPT = (
- "You are a master mechanic and service consultant. Extract a vehicle maintenance and repair graph.\n\n"
- "Guidelines:\n"
- "- Identify specific maintenance items and parts mentioned in the manual or log.\n"
- "- Map the conditions that trigger maintenance (e.g., Mileage, Warning Light, Visual Inspection).\n"
- "- Link parts to their respective systems and required actions."
-)
-
-MAINTENANCE_NODE_PROMPT = (
- "Extract maintenance tasks, parts, and fluid types. Include recommended intervals if specified."
-)
-
-MAINTENANCE_EDGE_PROMPT = (
- "Connect triggers to actions. Show which parts need replacement or inspection. Note specific precautions for each task."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class CarMaintenanceGraph(AutoGraph[MaintenanceItem, MaintenanceSequence]):
- """
- Template for vehicle service manuals, repair guides, and ownership logs.
-
- Helps organize maintenance schedules and part compatibility information.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = CarMaintenanceGraph(llm_client=llm, embedder=embedder)
- >>> text = "Change engine oil and filter every 10,000 km using synthetic 0W-20."
- >>> graph.feed_text(text)
- >>> print(graph.nodes) # Maintenance items
- """
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any
- ):
- """
- Initialize the CarMaintenanceGraph template.
-
- Args:
- llm_client: The language model client for extraction.
- embedder: The embedding model for deduplication.
- extraction_mode: "one_stage" or "two_stage".
- chunk_size: Max characters per chunk.
- chunk_overlap: Overlap between chunks.
- max_workers: Parallel processing workers.
- verbose: Enable progress logging.
- **kwargs: Extra arguments for AutoGraph.
- """
- super().__init__(
- node_schema=MaintenanceItem,
- edge_schema=MaintenanceSequence,
- node_key_extractor=lambda x: x.item_name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.action.lower()})-->{x.target.strip()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=MAINTENANCE_GRAPH_PROMPT,
- prompt_for_node_extraction=MAINTENANCE_NODE_PROMPT,
- prompt_for_edge_extraction=MAINTENANCE_EDGE_PROMPT,
- **kwargs
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: MaintenanceItem) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.item_name }{info}"
-
- def edge_label_extractor(edge: MaintenanceSequence) -> str:
- return f"{ edge.action }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/automotive/manufacturing.py b/hyperextract/templates/en/automotive/manufacturing.py
deleted file mode 100644
index 5304cdc..0000000
--- a/hyperextract/templates/en/automotive/manufacturing.py
+++ /dev/null
@@ -1,141 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class ManufacturingEntity(BaseModel):
- """An entity in the automotive supply chain or production line (Supplier, Factory, Assembly Step)."""
- name: str = Field(description="Name of the company, factory, or production process.")
- category: str = Field(
- description="Category: 'Supplier', 'Factory', 'Process', 'Material', 'Quality Check'."
- )
- specification: Optional[str] = Field(description="Production capacity, certifications, or lead times.")
-
-class SupplyChainRelation(BaseModel):
- """Connectivity between players and steps in automotive manufacturing."""
- source: str = Field(description="The upstream supplier or process.")
- target: str = Field(description="The downstream factory or product.")
- relation_type: str = Field(
- description="Type: 'Supplies to', 'Assembled at', 'Processed by', 'Certified for'."
- )
- logistics_details: Optional[str] = Field(description="Incoterms, transit times, or volume commitments.")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-MFG_GRAPH_PROMPT = (
- "You are a supply chain analyst for a major auto manufacturer. Extract the manufacturing and supply chain graph.\n\n"
- "Guidelines:\n"
- "- Identify the tiers of suppliers (Tier 1, Tier 2) and the components they provide.\n"
- "- Map the assembly sequence from raw materials to the final vehicle roll-out.\n"
- "- Capture geographic locations of factories and specific logistics constraints."
-)
-
-MFG_NODE_PROMPT = (
- "Extract suppliers, manufacturing plants, and critical production steps. Note their roles in the vehicle production cycle."
-)
-
-MFG_EDGE_PROMPT = (
- "Define the material and process flow. Connect suppliers to assembly plants and plants to the final car delivery."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class AutomotiveMfgGraph(AutoGraph[ManufacturingEntity, SupplyChainRelation]):
- """
- Template for automotive production lines, supply chain logistics, and supplier management.
-
- Helps visualize dependencies in the global automotive manufacturing ecosystem.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = AutomotiveMfgGraph(llm_client=llm, embedder=embedder)
- >>> text = "CATL supplies battery cells to the Tesla Gigafactory in Berlin for Model Y assembly."
- >>> graph.feed_text(text)
- >>> print(graph.edges) # Supply chain flow
- """
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any
- ):
- """
- Initialize the AutomotiveMfgGraph template.
-
- Args:
- llm_client: The language model client for extraction.
- embedder: The embedding model for deduplication.
- extraction_mode: "one_stage" or "two_stage".
- chunk_size: Max characters per chunk.
- chunk_overlap: Overlap between chunks.
- max_workers: Parallel processing workers.
- verbose: Enable progress logging.
- **kwargs: Extra arguments for AutoGraph.
- """
- super().__init__(
- node_schema=ManufacturingEntity,
- edge_schema=SupplyChainRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.strip()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=MFG_GRAPH_PROMPT,
- prompt_for_node_extraction=MFG_NODE_PROMPT,
- prompt_for_edge_extraction=MFG_EDGE_PROMPT,
- **kwargs
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: ManufacturingEntity) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.name }{info}"
-
- def edge_label_extractor(edge: SupplyChainRelation) -> str:
- return f"{ edge.relation_type }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/business/__init__.py b/hyperextract/templates/en/business/__init__.py
deleted file mode 100644
index 8da251c..0000000
--- a/hyperextract/templates/en/business/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .equity import EquityStructureGraph
-from .strategy import StrategicChainGraph
-from .ecosystem import MarketEcosystemHyper
-
-__all__ = [
- "EquityStructureGraph",
- "StrategicChainGraph",
- "MarketEcosystemHyper",
-]
diff --git a/hyperextract/templates/en/business/ecosystem.py b/hyperextract/templates/en/business/ecosystem.py
deleted file mode 100644
index 789d3ca..0000000
--- a/hyperextract/templates/en/business/ecosystem.py
+++ /dev/null
@@ -1,148 +0,0 @@
-from typing import List, Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoHypergraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class IndustryParticipant(BaseModel):
- """
- An entity (Company, Regulatory Body, NGO) within an industrial ecosystem.
- """
-
- name: str = Field(
- description="Name of the actor (e.g., 'TSMC', 'European Commission')."
- )
- rank: Optional[str] = Field(
- None,
- description="Market position: 'Tier 1 Supplier', 'Market Leader', 'Challenger'.",
- )
- technological_focus: Optional[str] = Field(
- None, description="Primary tech or domain (e.g., '5G', 'Lithography')."
- )
-
-
-class CollectiveAlliance(BaseModel):
- """
- A hyperedge representing a consortium, alliance, market cartel, or joint standard-setting body.
- """
-
- alliance_id: str = Field(
- description="Name or identifier of the alliance (e.g., 'CHIPS Alliance')."
- )
- participants: List[str] = Field(
- description="Names of all entities collaborating or competing in this group."
- )
- shared_goal: str = Field(
- description="The objective: 'Standardization', 'Anti-monopoly Review', 'Joint R&D'."
- )
- governance: Optional[str] = Field(
- None, description="Rules of participation or leading entity."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-ECOSYSTEM_CONSOLIDATED_PROMPT = (
- "You are a market strategist and antitrust expert. Extract market ecosystem dynamics.\n\n"
- "Rules:\n"
- "- Use Hyperedges to link multiple entities (3+) participating in a joint venture or alliance.\n"
- "- Capture non-binary group relations like cartels, industrial clusters, and regulatory coalitions.\n"
- "- Define the shared goal of these groups clearly."
-)
-
-ECOSYSTEM_NODE_PROMPT = (
- "You are a market analyst. Your task is to identify all participants (Nodes) within an industrial ecosystem.\n\n"
- "Extraction Rules:\n"
- "- Identify companies, NGOs, and regulatory bodies mentioned.\n"
- "- Capture their market rank and primary technological focus area.\n"
- "- Focus on naming actors that collaborate or compete as a group.\n"
- "- DO NOT identify alliances, joint ventures, or consortia at this stage."
-)
-
-ECOSYSTEM_EDGE_PROMPT = (
- "You are a market analyst. Given the industrial participants, extract the collective relationships (Hyperedges).\n\n"
- "Extraction Rules:\n"
- "- Group multiple participants (3+) into alliances, standard bodies, or joint ventures.\n"
- "- Define the shared goal (e.g., 'R&D', 'Anti-monopoly Review') for the entire group.\n"
- "- Record any mentioned governance structures or leading participants.\n"
- "- Ensure every group member is selected from the provided participants list."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class MarketEcosystemHyper(AutoHypergraph[IndustryParticipant, CollectiveAlliance]):
- """
- Hypergraph template for industrial clusters, ecosystem coalitions, and complex market alliances.
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- super().__init__(
- node_schema=IndustryParticipant,
- edge_schema=CollectiveAlliance,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: x.alliance_id.strip(),
- nodes_in_edge_extractor=lambda x: tuple(x.participants),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=ECOSYSTEM_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=ECOSYSTEM_NODE_PROMPT,
- prompt_for_edge_extraction=ECOSYSTEM_EDGE_PROMPT,
- **kwargs,
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: IndustryParticipant) -> str:
- return f"{ node.name }"
-
- def edge_label_extractor(edge: CollectiveAlliance) -> str:
- return f"{ edge.alliance_id }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/business/equity.py b/hyperextract/templates/en/business/equity.py
deleted file mode 100644
index 84109b0..0000000
--- a/hyperextract/templates/en/business/equity.py
+++ /dev/null
@@ -1,156 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class StakeholderNode(BaseModel):
- """
- A legal or natural person holding equity or control in a business.
- """
-
- name: str = Field(
- description="Full registry name of the company or individual's full name."
- )
- type: str = Field(
- description="Type: 'Natural Person', 'Public Company', 'Private Company', 'State-Owned Enterprise', 'Venture Capital'."
- )
- jurisdiction: Optional[str] = Field(
- None, description="Country or region of incorporation/residence."
- )
- is_ubo: bool = Field(
- False,
- description="Whether this entity is identified as an Ultimate Beneficial Owner.",
- )
-
-
-class OwnershipRelation(BaseModel):
- """
- A stakeholding or control link between two entities.
- """
-
- source: str = Field(description="The shareholder or parent entity.")
- target: str = Field(
- description="The company being owned or controlled (subsidiary)."
- )
- stake_percentage: Optional[float] = Field(
- None, description="Percentage of ownership (0.0 to 100.0)."
- )
- control_type: str = Field(
- "Equity",
- description="Nature of link: 'Equity Acquisition', 'Voting Rights', 'Nominee Shareholding', 'Direct Control'.",
- )
- as_of_date: Optional[str] = Field(
- None, description="The date or fiscal period when this ownership was recorded."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-EQUITY_CONSOLIDATED_PROMPT = (
- "You are a corporate paralegal and financial analyst. Extract complex ownership structures.\n\n"
- "Rules:\n"
- "- Prioritize identifying the percentage of shares held by each entity.\n"
- "- Distinguish between direct shareholders and Ultimate Beneficial Owners (UBO).\n"
- "- Map the hierarchy from the parent organization down to the smallest subsidiary."
-)
-
-EQUITY_NODE_PROMPT = (
- "You are a corporate paralegal and financial analyst. Your task is to identify all stakeholders (Nodes) within a corporate structure.\n\n"
- "Extraction Rules:\n"
- "- Identify companies, investment funds, government bodies, and individual persons.\n"
- "- Classify each as 'Natural Person', 'Public Company', 'Private Company', etc.\n"
- "- Flag entities identified as Ultimate Beneficial Owners (UBO) where explicit.\n"
- "- DO NOT extract ownership percentages or control links at this stage."
-)
-
-EQUITY_EDGE_PROMPT = (
- "You are a corporate paralegal and financial analyst. Given the list of stakeholders, map the ownership and control relationships (Edges).\n\n"
- "Extraction Rules:\n"
- "- Extract the exact percentage of equity held between a shareholder (source) and a company (target).\n"
- "- Specify the nature of control (e.g., Voting Rights, Direct Control, Nominee Shareholding).\n"
- "- Create edges only between entities found in the provided stakeholder list.\n"
- "- Capture the 'as of' date or historical period for the ownership record."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class EquityStructureGraph(AutoGraph[StakeholderNode, OwnershipRelation]):
- """
- Precision template for corporate registries, M&A analysis, and KYC/UBO tracking.
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- super().__init__(
- node_schema=StakeholderNode,
- edge_schema=OwnershipRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}->{x.target.strip()}_at_{x.stake_percentage}%"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=EQUITY_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=EQUITY_NODE_PROMPT,
- prompt_for_edge_extraction=EQUITY_EDGE_PROMPT,
- **kwargs,
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: StakeholderNode) -> str:
- info = f" ({ node.type })" if getattr(node, "type", None) else ""
- return f"{ node.name }{info}"
-
- def edge_label_extractor(edge: OwnershipRelation) -> str:
- return f"{edge.control_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/business/strategy.py b/hyperextract/templates/en/business/strategy.py
deleted file mode 100644
index 8171d6a..0000000
--- a/hyperextract/templates/en/business/strategy.py
+++ /dev/null
@@ -1,153 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoTemporalGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class StrategicNode(BaseModel):
- """
- An organizational entity, business division, or strategic objective.
- """
-
- name: str = Field(
- description="Name of the strategic unit or goal (e.g., 'Azure', 'Carbon Neutrality')."
- )
- scope: str = Field(
- description="Scope: 'Global', 'Regional', 'Product-specific', 'Corporate-wide'."
- )
- importance: Optional[str] = Field(
- None, description="Criticality: 'Core Business', 'Emerging Market', 'Legacy'."
- )
-
-
-class StrategicPivot(BaseModel):
- """
- A temporal transition representing a change in strategic focus, investment, or divestiture.
- """
-
- source: str = Field(description="The original area of focus or acting entity.")
- target: str = Field(description="The new target, direction, or goal.")
- action: str = Field(
- description="Strategic action: 'Pivot to', 'Acquire', 'Divest', 'Merge', 'Sunset', 'Scale'."
- )
- timestamp: str = Field(
- description="When the shift occurred (e.g., 'Q4 2022', '2025-01-01')."
- )
- rationale: Optional[str] = Field(
- None,
- description="The 'Why' behind the move (market pressure, CEO change, tech shift).",
- )
- budget: Optional[str] = Field(
- None, description="Financial commitment mentioned alongside the pivot."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-STRATEGY_CONSOLIDATED_PROMPT = (
- "You are a management consultant and industry analyst. Extract the evolution of corporate strategy.\n\n"
- "Rules:\n"
- "- Focus on 'Pivots': movements from one business area to another.\n"
- "- Link specific actions to concrete timestamps.\n"
- "- Capture the underlying reasoning (rationale) for each strategic change."
-)
-
-STRATEGY_NODE_PROMPT = (
- "You are a management consultant. Your task is to identify strategic entities and objectives (Nodes).\n\n"
- "Extraction Rules:\n"
- "- Identify business divisions, specific products, and overarching corporate goals (e.g., 'Sustainability').\n"
- "- Determine the scope and organizational importance of each unit.\n"
- "- Focus on nouns and noun phrases representing strategic focus areas.\n"
- "- DO NOT extract strategic moves or transitions at this stage."
-)
-
-STRATEGY_EDGE_PROMPT = (
- "You are a management consultant. Given the identified strategic units, map the temporal shifts in strategy (Edges).\n\n"
- "Extraction Rules:\n"
- "- Extract 'Pivots' and actions like 'Acquire', 'Divest', or 'Sunset'.\n"
- "- Link the source (old focus) to the target (new focus) with a precise timestamp.\n"
- "- Capture the underlying rationale and any financial commitment (budget) mentioned.\n"
- "- Only reference nodes present in the strategic units list."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class StrategicChainGraph(AutoTemporalGraph[StrategicNode, StrategicPivot]):
- """
- Temporal template for tracking corporate strategy, longitudinal organizational change, and market shifts.
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- super().__init__(
- node_schema=StrategicNode,
- edge_schema=StrategicPivot,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}-({x.action})->{x.target.strip()}"
- ),
- time_in_edge_extractor=lambda x: x.timestamp.strip(),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=STRATEGY_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=STRATEGY_NODE_PROMPT,
- prompt_for_edge_extraction=STRATEGY_EDGE_PROMPT,
- **kwargs,
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: StrategicNode) -> str:
- return f"{node.name}"
-
- def edge_label_extractor(edge: StrategicPivot) -> str:
- return f"{edge.action}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/construction/__init__.py b/hyperextract/templates/en/construction/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/construction/project_timeline.py b/hyperextract/templates/en/construction/project_timeline.py
deleted file mode 100644
index 247086a..0000000
--- a/hyperextract/templates/en/construction/project_timeline.py
+++ /dev/null
@@ -1,185 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoTemporalGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class MilestoneTaskNode(BaseModel):
- """
- A specific task, milestone, or phase in a construction project schedule.
- """
-
- task_name: str = Field(
- description="Operational name of the task (e.g., 'Foundation Pouring')."
- )
- phase: str = Field(
- description="Project phase: 'Pre-con', 'Substructure', 'Superstructure', 'Fit-out', 'Commissioning'."
- )
- duration_estimate: Optional[str] = Field(
- None, description="Estimated time required (e.g., '14 days', '2 months')."
- )
- responsible_party: Optional[str] = Field(
- None, description="Subcontractor or team lead in charge."
- )
-
-
-class TemporalDependency(BaseModel):
- """
- A time-based link indicating task sequencing.
- """
-
- source: str = Field(description="The predecessor task.")
- target: str = Field(description="The successor task.")
- timestamp: str = Field(
- description="Temporal marker or sequence ID (e.g., 'Phase 1', 'Step 2', or a specific date/time)."
- )
- dependency_type: str = Field(
- "Finish-to-Start",
- description="Type: 'Finish-to-Start (FS)', 'Start-to-Start (SS)', 'Finish-to-Finish (FF)', 'Start-to-Finish (SF)'.",
- )
- lag_time: Optional[str] = Field(
- None,
- description="Wait time between tasks (e.g., '3 days for concrete curing').",
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CONSTRUCTION_TEMPORAL_CONSOLIDATED_PROMPT = (
- "You are a Senior Project Planner and Scheduler (P6/MS Project expert). Extract the construction project timeline.\n\n"
- "Rules:\n"
- "- Identify granular tasks and major milestones as Nodes.\n"
- "- Establish clear temporal dependencies (Edges) using industry-standard logic (FS, SS, etc.).\n"
- "- Capture durations and curing/lag times where mentioned.\n"
- "- Focus on the critical path and sequencing logic."
-)
-
-CONSTRUCTION_TEMPORAL_NODE_PROMPT = (
- "Identify all tasks, scheduled activities, and project milestones.\n\n"
- "Rules:\n"
- "- Extract action-oriented task names.\n"
- "- Group tasks into project phases and identify responsible parties.\n"
- "- DO NOT identify sequences or dependencies yet."
-)
-
-CONSTRUCTION_TEMPORAL_EDGE_PROMPT = (
- "Link construction tasks based on their logical sequence in time.\n\n"
- "Rules:\n"
- "- Define the predecessor and successor for each link.\n"
- "- Classify the link type (FS is default if not specified).\n"
- "- Identify any required lag times (e.g., waiting for inspection or drying).\n"
- "- Only link tasks from the provided list."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class ProjectSchedulingTimeline(
- AutoTemporalGraph[MilestoneTaskNode, TemporalDependency]
-):
- """
- Template for extracting project schedules and Critical Path Method (CPM) networks.
-
- Ideal for schedule risk analysis, delay tracking, and progress monitoring.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> timeline = ProjectSchedulingTimeline(llm_client=llm, embedder=embedder)
- >>> timeline.feed_text("Excavation must finish before the foundation slab can be poured.")
- >>> print(timeline.nodes) # Access tasks and milestones
- >>> print(timeline.edges) # Access temporal dependencies
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the Project Scheduling Timeline template.
-
- Args:
- llm_client (BaseChatModel): The language model client used for schedule extraction.
- embedder (Embeddings): The embedding model used for task and dependency deduplication.
- extraction_mode (str, optional): 'one_stage' for joint extraction,
- 'two_stage' for separate passes. Defaults to "one_stage".
- chunk_size (int, optional): Maximum characters per text chunk. Defaults to 2048.
- chunk_overlap (int, optional): Overlap between chunks. Defaults to 256.
- max_workers (int, optional): Parallel processing workers. Defaults to 10.
- verbose (bool, optional): If True, enables progress logging. Defaults to False.
- **kwargs (Any): Additional parameters for the AutoTemporalGraph base class.
- """
- super().__init__(
- node_schema=MilestoneTaskNode,
- edge_schema=TemporalDependency,
- node_key_extractor=lambda x: x.task_name.strip().lower(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip().lower()}--[{x.dependency_type}]-->{x.target.strip().lower()}"
- ),
- nodes_in_edge_extractor=lambda x: (
- x.source.strip().lower(),
- x.target.strip().lower(),
- ),
- time_in_edge_extractor=lambda x: x.timestamp.strip(),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CONSTRUCTION_TEMPORAL_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CONSTRUCTION_TEMPORAL_NODE_PROMPT,
- prompt_for_edge_extraction=CONSTRUCTION_TEMPORAL_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: MilestoneTaskNode) -> str:
- info = f" ({node.phase})" if getattr(node, "phase", None) else ""
- return f"{node.task_name}{info}"
-
- def edge_label_extractor(edge: TemporalDependency) -> str:
- return f"{edge.dependency_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/construction/resource_registry.py b/hyperextract/templates/en/construction/resource_registry.py
deleted file mode 100644
index 8d23f05..0000000
--- a/hyperextract/templates/en/construction/resource_registry.py
+++ /dev/null
@@ -1,108 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoSet
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class ConstructionResource(BaseModel):
- """
- A specific asset, material, or labor category assigned to a construction project.
- """
- resource_name: str = Field(description="Name of the resource (e.g., 'Tower Crane TC-01', 'Fly Ash Concrete').")
- category: str = Field(description="Category: 'Heavy Machinery', 'Bulk Material', 'Skilled Labor', 'Fixed Equipment', 'Consumables'.")
- quantity: Optional[str] = Field(None, description="Current quantity or capacity on site.")
- supplier_or_team: Optional[str] = Field(None, description="The entity providing or managing this resource.")
- status: str = Field("Active", description="Current operational status: 'Active', 'Idle', 'Maintenance', 'Depleted'.")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CONSTRUCTION_RESOURCE_PROMPT = (
- "You are a Construction Logistics and Procurement Manager. Extract all project resources and assets.\n\n"
- "Rules:\n"
- "- Identify machinery, materials, and labor types.\n"
- "- Capture specific quantities and supplier names if mentioned.\n"
- "- Deduplicate similar items (e.g., 'Excavator A' and 'The primary excavator' should be merged if referring to the same unit)."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class ResourceRegistrySet(AutoSet[ConstructionResource]):
- """
- Template for managing a consolidated site asset and material registry.
-
- Uses AutoSet for automatic deduplication and merging of multi-chunk resource mentions.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> registry = ResourceRegistrySet(llm_client=llm, embedder=embedder)
- >>> registry.feed_text("We have two 50-ton cranes from Liebherr on site.")
- >>> registry.feed_text("The Liebherr cranes are currently being serviced.")
- >>> print(registry.items) # Mentions will be merged into a single Tower Crane resource.
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the Resource Registry Set template.
-
- Args:
- llm_client (BaseChatModel): The language model client used for resource extraction.
- embedder (Embeddings): The embedding model used for resource deduplication.
- chunk_size (int, optional): Maximum characters per text chunk. Defaults to 2048.
- chunk_overlap (int, optional): Overlap between chunks. Defaults to 256.
- max_workers (int, optional): Parallel processing workers. Defaults to 10.
- verbose (bool, optional): If True, enables progress logging. Defaults to False.
- **kwargs (Any): Additional parameters for the AutoSet base class.
- """
- super().__init__(
- item_schema=ConstructionResource,
- key_extractor=lambda x: x.resource_name.strip().lower(),
- llm_client=llm_client,
- embedder=embedder,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CONSTRUCTION_RESOURCE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_for_search: int = 3,
- top_k_for_chat: int = 3,
- ) -> None:
- """
- Visualize the collection using OntoSight.
-
- Args:
- top_k_for_search (int): Number of items to retrieve for search context. Default 3.
- top_k_for_chat (int): Number of items to retrieve for chat context. Default 3.
- """
- def item_label_extractor(item: ConstructionResource) -> str:
- info = f" ({ item.category })" if getattr(item, "category", None) else ""
- return f"{ item.resource_name }{info}"
-
- super().show(
- item_label_extractor=item_label_extractor,
- top_k_for_search=top_k_for_search,
- top_k_for_chat=top_k_for_chat,
- )
diff --git a/hyperextract/templates/en/construction/site_topology.py b/hyperextract/templates/en/construction/site_topology.py
deleted file mode 100644
index b17b294..0000000
--- a/hyperextract/templates/en/construction/site_topology.py
+++ /dev/null
@@ -1,180 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoSpatialGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class PhysicalElementNode(BaseModel):
- """
- A specific physical component or zone within a construction site.
- """
-
- name: str = Field(
- description="Unique identifier or name of the element (e.g., 'Column-A1', 'Zone-B')."
- )
- element_type: str = Field(
- description="Type: 'Structural', 'MEP', 'Architectural', 'Temporary Structure', 'Safety Zone'."
- )
- location_coordinates: Optional[str] = Field(
- None,
- description="Spatial coordinates or relative grid location (e.g., 'X:10, Y:20, Z:5').",
- )
- material_properties: Optional[str] = Field(
- None, description="Key material info (e.g., 'C30 Concrete', 'Steel S355')."
- )
-
-
-class SpatialRelation(BaseModel):
- """
- A spatial or topological connection between two physical elements.
- """
-
- source: str = Field(description="The reference element.")
- target: str = Field(description="The related element.")
- location_marker: str = Field(
- description="The physical location or zone where this relationship resides (e.g., 'Level 3', 'North Wing', 'Grid A-1')."
- )
- relation_type: str = Field(
- description="Type: 'Supporting', 'Adjacent', 'Enclosed By', 'Intersects', 'Aligned With'."
- )
- offset_distance: Optional[float] = Field(
- None, description="Physical distance between elements in meters."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CONSTRUCTION_SPATIAL_CONSOLIDATED_PROMPT = (
- "You are a Senior BIM (Building Information Modeling) Engineer and Site Surveyor. Extract the spatial topology of the construction project.\n\n"
- "Rules:\n"
- "- Identify fixed physical elements and their properties as Nodes.\n"
- "- Map the structural and spatial relationships (Edges) between these elements.\n"
- "- Ensure coordinates and distances are captured if mentioned.\n"
- "- Focus on how objects are physically connected or positioned relative to each other."
-)
-
-CONSTRUCTION_SPATIAL_NODE_PROMPT = (
- "Identify all physical building components and site zones.\n\n"
- "Rules:\n"
- "- Extract distinct structural, mechanical, or architectural elements.\n"
- "- Capture any mentioned locations, levels, or material information.\n"
- "- DO NOT identify connections at this stage."
-)
-
-CONSTRUCTION_SPATIAL_EDGE_PROMPT = (
- "Map the spatial and structural relationships between the identified building elements.\n\n"
- "Rules:\n"
- "- Determine how Element A relates to Element B (e.g., A supports B, A is adjacent to B).\n"
- "- Extract physical offsets or distances if provided.\n"
- "- Only link elements from the provided list."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class SiteTopologyGraph(AutoSpatialGraph[PhysicalElementNode, SpatialRelation]):
- """
- Template for building a spatial digital twin of a construction site.
-
- Useful for structural analysis, BIM-to-Graph conversion, and site logistics planning.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> graph = SiteTopologyGraph(llm_client=llm, embedder=embedder)
- >>> graph.feed_text("The foundation slab supports the four main corner columns.")
- >>> print(graph.nodes) # Access structural elements
- >>> print(graph.edges) # Access spatial relationships
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the Site Topology Graph template.
-
- Args:
- llm_client (BaseChatModel): The language model client used for topology extraction.
- embedder (Embeddings): The embedding model used for element deduplication.
- extraction_mode (str, optional): 'one_stage' for joint extraction,
- 'two_stage' for separate passes. Defaults to "one_stage".
- chunk_size (int, optional): Maximum characters per text chunk. Defaults to 2048.
- chunk_overlap (int, optional): Overlap between chunks. Defaults to 256.
- max_workers (int, optional): Parallel processing workers. Defaults to 10.
- verbose (bool, optional): If True, enables progress logging. Defaults to False.
- **kwargs (Any): Additional parameters for the AutoSpatialGraph base class.
- """
- super().__init__(
- node_schema=PhysicalElementNode,
- edge_schema=SpatialRelation,
- node_key_extractor=lambda x: x.name.strip().lower(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip().lower()}--{x.relation_type.lower()}--{x.target.strip().lower()}"
- ),
- nodes_in_edge_extractor=lambda x: (
- x.source.strip().lower(),
- x.target.strip().lower(),
- ),
- location_in_edge_extractor=lambda x: x.location_marker.strip(),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CONSTRUCTION_SPATIAL_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CONSTRUCTION_SPATIAL_NODE_PROMPT,
- prompt_for_edge_extraction=CONSTRUCTION_SPATIAL_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: PhysicalElementNode) -> str:
- return f"{node.name}"
-
- def edge_label_extractor(edge: SpatialRelation) -> str:
- return f"{edge.relation_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/cybersecurity/__init__.py b/hyperextract/templates/en/cybersecurity/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/cybersecurity/intrusion_kill_chain.py b/hyperextract/templates/en/cybersecurity/intrusion_kill_chain.py
deleted file mode 100644
index d69cf63..0000000
--- a/hyperextract/templates/en/cybersecurity/intrusion_kill_chain.py
+++ /dev/null
@@ -1,179 +0,0 @@
-from typing import Optional, List, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class AttackStageNode(BaseModel):
- """
- A specific phase or action in a cyber attack (aligned with MITRE ATT&CK or Lockheed Kill Chain).
- """
-
- stage_name: str = Field(
- description="Name of the attack stage (e.g., 'Initial Access', 'Lateral Movement')."
- )
- tactics_used: List[str] = Field(
- default_factory=list,
- description="Specific techniques or tactics employed (e.g., 'Phishing', 'Pass-the-Hash').",
- )
- adversary_tools: List[str] = Field(
- default_factory=list,
- description="Malware or tools used in this stage (e.g., 'Cobalt Strike', 'Mimikatz').",
- )
- timestamp: Optional[str] = Field(
- None, description="Time or sequence marker for the action."
- )
-
-
-class KillChainLink(BaseModel):
- """
- A directional link representing the progression of an attack from one stage to another.
- """
-
- source: str = Field(description="The preceding attack stage.")
- target: str = Field(description="The succeeding attack stage.")
- causality: str = Field(
- "Progression",
- description="Type: 'Progression' (standard), 'Triggered' (automatic), 'Pivoted' (infrastructure shift).",
- )
- observables: List[str] = Field(
- default_factory=list,
- description="IOCs (Indicators of Compromise) linking the stages.",
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CYBER_KILL_CHAIN_CONSOLIDATED_PROMPT = (
- "You are a Forensic Investigator and Incident Responder. Reconstruct the attack timeline and kill chain from incident reports.\n\n"
- "Rules:\n"
- "- Identify distinct attack stages as Nodes.\n"
- "- Map the chronological and logical progression (Edges) of the intrusion.\n"
- "- Capture specific tactics, tools (TTPs), and timestamps for each stage.\n"
- "- Focus on the 'how' and 'when' of the adversary's movement."
-)
-
-CYBER_KILL_CHAIN_NODE_PROMPT = (
- "Identify all suspicious activities and distinct phases of a cyber intrusion.\n\n"
- "Rules:\n"
- "- Extract action-oriented stages (e.g., 'Delivery of Payload', 'C2 Establishment').\n"
- "- Identify the tools and techniques associated with each discovery.\n"
- "- DO NOT link the stages yet."
-)
-
-CYBER_KILL_CHAIN_EDGE_PROMPT = (
- "Establish the sequence and progression between identified attack stages.\n\n"
- "Rules:\n"
- "- Define which stage led to or enabled the next.\n"
- "- Identify Indicators of Compromise (IPs, Hashes) that connect the stages.\n"
- "- Only link stages from the provided list."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class IntrusionKillChainGraph(AutoGraph[AttackStageNode, KillChainLink]):
- """
- Template for reconstructing cyber attack lifecycles.
-
- Useful for incident response (IR), forensic reporting, and threat hunting.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> kill_chain = IntrusionKillChainGraph(llm_client=llm, embedder=embedder)
- >>> kill_chain.feed_text("The attacker gained initial access via a phishing email.")
- >>> print(kill_chain.nodes) # Access attack stages
- >>> print(kill_chain.edges) # Access progression links
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the Intrusion Kill Chain Graph template.
-
- Args:
- llm_client (BaseChatModel): The language model client used for kill chain extraction.
- embedder (Embeddings): The embedding model used for stage deduplication.
- extraction_mode (str, optional): 'one_stage' for joint extraction,
- 'two_stage' for separate passes. Defaults to "one_stage".
- chunk_size (int, optional): Maximum characters per text chunk. Defaults to 2048.
- chunk_overlap (int, optional): Overlap between chunks. Defaults to 256.
- max_workers (int, optional): Parallel processing workers. Defaults to 10.
- verbose (bool, optional): If True, enables progress logging. Defaults to False.
- **kwargs (Any): Additional parameters for the AutoGraph base class.
- """
- super().__init__(
- node_schema=AttackStageNode,
- edge_schema=KillChainLink,
- node_key_extractor=lambda x: x.stage_name.strip().lower(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip().lower()}->{x.target.strip().lower()}"
- ),
- nodes_in_edge_extractor=lambda x: (
- x.source.strip().lower(),
- x.target.strip().lower(),
- ),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CYBER_KILL_CHAIN_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CYBER_KILL_CHAIN_NODE_PROMPT,
- prompt_for_edge_extraction=CYBER_KILL_CHAIN_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: AttackStageNode) -> str:
- return f"{node.stage_name}"
-
- def edge_label_extractor(edge: KillChainLink) -> str:
- return f"{edge.causality}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/cybersecurity/threat_intelligence_nexus.py b/hyperextract/templates/en/cybersecurity/threat_intelligence_nexus.py
deleted file mode 100644
index 2c28541..0000000
--- a/hyperextract/templates/en/cybersecurity/threat_intelligence_nexus.py
+++ /dev/null
@@ -1,181 +0,0 @@
-from typing import List, Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoHypergraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class ThreatActorNode(BaseModel):
- """
- An entity (individual, group, or nation-state) responsible for cyber threats.
- """
-
- actor_name: str = Field(
- description="Alias or group name (e.g., 'APT28', 'Fancy Bear', 'Lazarus Group')."
- )
- origin_country: Optional[str] = Field(
- None, description="Suspected geographic origin."
- )
- motivation: Optional[str] = Field(
- None,
- description="Primary driver: 'Espionage', 'Financial Gain', 'Sabotage', 'Hacktivism'.",
- )
- typical_targets: List[str] = Field(
- default_factory=list, description="Commonly targeted industries or regions."
- )
-
-
-class IntelligenceNexus(BaseModel):
- """
- A hyperedge connecting an actor, specific infrastructure, and a campaign theme.
- """
-
- participants: List[str] = Field(
- description="Entities involved: Actor names, Tool names, Infrastructure (IPs/Domains), Target Organizations."
- )
- campaign_name: str = Field(
- description="Known name of the operation (e.g., 'SolarWinds Supply Chain Attack')."
- )
- shared_ttp: str = Field(
- description="The common tactic, technique, or procedure that links these disparate entities."
- )
- confidence_level: str = Field(
- "Medium", description="Analytical confidence: 'High', 'Medium', 'Low'."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CYBER_NEXUS_CONSOLIDATED_PROMPT = (
- "You are a Strategic Threat Intelligence Analyst. Map the high-level correlations between actors, infrastructure, and infrastructure.\n\n"
- "Rules:\n"
- "- Identify threat actors and their aliases as Nodes.\n"
- "- Create Hyperedges (Nexuses) that link actors to specific tools and infrastructure under a named campaign.\n"
- "- Focus on multi-point connections that reveal a cohesive operation rather than simple binary links.\n"
- "- Highlight shared TTPs that unify the nexus."
-)
-
-CYBER_NEXUS_NODE_PROMPT = (
- "Identify threat actors, malware families, and malicious infrastructure entities.\n\n"
- "Rules:\n"
- "- Extract actor aliases and group names.\n"
- "- Identify motivations and target industries.\n"
- "- DO NOT identify campaigns or relations yet."
-)
-
-CYBER_NEXUS_EDGE_PROMPT = (
- "Identify the operational Campaigns or Nexuses that unify actors, tools, and targets.\n\n"
- "Rules:\n"
- "- Each hyperedge must describe a single campaign or operation.\n"
- "- Link the actor(s) to the specific infrastructure and tools used in that operation.\n"
- "- Only link entities from the provided list."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class ThreatIntelligenceNexusHypergraph(
- AutoHypergraph[ThreatActorNode, IntelligenceNexus]
-):
- """
- Template for strategic threat intelligence mapping.
-
- Ideal for connecting dots between infrastructure, actors, and malware in complex campaigns.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> nexus = ThreatIntelligenceNexusHypergraph(llm_client=llm, embedder=embedder)
- >>> nexus.feed_text("APT28 utilized X-Tunnel and C2 at 1.2.3.4 during Operation Fancy-Phish.")
- >>> print(nexus.nodes) # Access threat actors
- >>> print(nexus.edges) # Access campaign hyperedges
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the Threat Intelligence Nexus Hypergraph template.
-
- Args:
- llm_client (BaseChatModel): The language model client used for nexus extraction.
- embedder (Embeddings): The embedding model used for actor deduplication.
- extraction_mode (str, optional): 'one_stage' for joint extraction,
- 'two_stage' for separate passes. Defaults to "one_stage".
- chunk_size (int, optional): Maximum characters per text chunk. Defaults to 2048.
- chunk_overlap (int, optional): Overlap between chunks. Defaults to 256.
- max_workers (int, optional): Parallel processing workers. Defaults to 10.
- verbose (bool, optional): If True, enables progress logging. Defaults to False.
- **kwargs (Any): Additional parameters for the AutoHypergraph base class.
- """
- super().__init__(
- node_schema=ThreatActorNode,
- edge_schema=IntelligenceNexus,
- node_key_extractor=lambda x: x.actor_name.strip().lower(),
- edge_key_extractor=lambda x: (
- f"{x.campaign_name.strip().lower()}_linking_{sorted(x.participants)}"
- ),
- nodes_in_edge_extractor=lambda x: [
- p.strip().lower() for p in x.participants
- ],
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CYBER_NEXUS_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CYBER_NEXUS_NODE_PROMPT,
- prompt_for_edge_extraction=CYBER_NEXUS_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: ThreatActorNode) -> str:
- return f"{node.actor_name}"
-
- def edge_label_extractor(edge: IntelligenceNexus) -> str:
- return f"{edge.campaign_name}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/en/cybersecurity/vulnerability_catalog.py b/hyperextract/templates/en/cybersecurity/vulnerability_catalog.py
deleted file mode 100644
index cbb6616..0000000
--- a/hyperextract/templates/en/cybersecurity/vulnerability_catalog.py
+++ /dev/null
@@ -1,123 +0,0 @@
-from typing import Optional, List, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoSet
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class VulnerabilityEntry(BaseModel):
- """
- A specific security vulnerability or weakness identified in a system or software.
- """
-
- vulnerability_id: str = Field(
- description="Unique identifier (e.g., CVE-2023-1234, CWE-79)."
- )
- title: str = Field(description="Short descriptive name of the vulnerability.")
- severity: str = Field(
- description="Severity rating: 'Critical', 'High', 'Medium', 'Low', 'Informational'."
- )
- cvss_score: Optional[float] = Field(
- None, description="CVSS base score (0.0 to 10.0)."
- )
- affected_components: List[str] = Field(
- default_factory=list, description="Software or hardware components impacted."
- )
- remediation: Optional[str] = Field(
- None, description="Recommended fix or mitigation steps."
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CYBER_VULN_PROMPT = (
- "You are a Senior Security Analyst and Vulnerability Researcher. Extract and consolidate security vulnerabilities from threat reports, advisories, or scan results.\n\n"
- "Rules:\n"
- "- Identify unique vulnerabilities by their CVE/CWE ID or title.\n"
- "- Capture severity levels and CVSS scores accurately.\n"
- "- Merge duplicate mentions of the same vulnerability into a single entry, combining information about affected components and remediation."
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class VulnerabilityUnifiedCatalog(AutoSet[VulnerabilityEntry]):
- """
- Template for building a deduplicated catalog of security vulnerabilities.
-
- Ideal for processing multiple security advisories and creating a master list for patching.
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> catalog = VulnerabilityUnifiedCatalog(llm_client=llm, embedder=embedder)
- >>> catalog.feed_text("CVE-2024-0001 is a critical RCE in the kernel.")
- >>> catalog.feed_text("The kernel RCE (CVE-2024-0001) requires an immediate patch.")
- >>> print(catalog.items) # Duplicate entries for CVE-2024-0001 will be merged
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- Initialize the Vulnerability Unified Catalog template.
-
- Args:
- llm_client (BaseChatModel): The language model client used for vulnerability extraction.
- embedder (Embeddings): The embedding model used for entry deduplication.
- chunk_size (int, optional): Maximum characters per text chunk. Defaults to 2048.
- chunk_overlap (int, optional): Overlap between chunks. Defaults to 256.
- max_workers (int, optional): Parallel processing workers. Defaults to 10.
- verbose (bool, optional): If True, enables progress logging. Defaults to False.
- **kwargs (Any): Additional parameters for the AutoSet base class.
- """
- super().__init__(
- item_schema=VulnerabilityEntry,
- key_extractor=lambda x: x.vulnerability_id.strip().upper(),
- llm_client=llm_client,
- embedder=embedder,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CYBER_VULN_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_for_search: int = 3,
- top_k_for_chat: int = 3,
- ) -> None:
- """
- Visualize the collection using OntoSight.
-
- Args:
- top_k_for_search (int): Number of items to retrieve for search context. Default 3.
- top_k_for_chat (int): Number of items to retrieve for chat context. Default 3.
- """
-
- def item_label_extractor(item: VulnerabilityEntry) -> str:
- return f"{item.title}"
-
- super().show(
- item_label_extractor=item_label_extractor,
- top_k_for_search=top_k_for_search,
- top_k_for_chat=top_k_for_chat,
- )
diff --git a/hyperextract/templates/en/fashion/__init__.py b/hyperextract/templates/en/fashion/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/marketing/__init__.py b/hyperextract/templates/en/marketing/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/music/__init__.py b/hyperextract/templates/en/music/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/politics/__init__.py b/hyperextract/templates/en/politics/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/real_estate/__init__.py b/hyperextract/templates/en/real_estate/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/recruitment/__init__.py b/hyperextract/templates/en/recruitment/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/software/__init__.py b/hyperextract/templates/en/software/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/sports/__init__.py b/hyperextract/templates/en/sports/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/supply_chain/__init__.py b/hyperextract/templates/en/supply_chain/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/en/travel/__init__.py b/hyperextract/templates/en/travel/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/README.md b/hyperextract/templates/zh/README.md
index c634969..2c4ce32 100644
--- a/hyperextract/templates/zh/README.md
+++ b/hyperextract/templates/zh/README.md
@@ -11,41 +11,26 @@
| 目录名称 | 说明 | 应用场景示例 |
| :--- | :--- | :--- |
-| **`tcm`** | **中医 (特色)** | 中草药方剂、经络穴位、辨证施治过程等。 |
-| **`wuxia`** | **武侠/修真 (特色)** | 门派恩怨、功法境界、修仙等级体系、法宝神兵。 |
| **`general`** | 通用百科 | 人物、机构、事件、地点等基础实体的快速抽取。 |
| **`finance`** | 金融投资 | A股/港股研报、基金招募说明书、上市公司季报。 |
| **`legal`** | 法律合规 | 合同关键要素(义务、权利)、案情调查、法条引用。 |
| **`education`** | 教育学术 | K12知识点图谱、课程大纲依赖关系、学术论文脉络。 |
-| **`business`** | 商业通用 | 公司全画像、股权穿透结构、商业战略目标。 |
| **`medicine`** | 医学健康 | 疾病症状映射、临床诊疗路径、西药说明书结构化。 |
| **`food`** | 美食餐饮 | 中餐菜谱(刀工、火候)、食材配比、八大菜系特征。 |
| **`history`** | 历史文化 | 朝代更迭时间线、重大战役、历史人物生平关系。 |
| **`literature`** | 文学 | 小说人物关系网、文学流派演变、古诗词意象关联。 |
-| **`software`** | 软件与IT | 系统架构图、接口定义、代码依赖关系、DevOps流程。 |
| **`game`** | 游戏娱乐 | 游戏世界观设定、任务流向树、物品合成与掉落表。 |
-| **`sports`** | 体育竞技 | 赛事文字直播提取、球员身价与转会网络、战术阵型。 |
-| **`travel`** | 旅游出行 | 游记攻略结构化、景点关系图、旅行行程规划。 |
-| **`recruitment`**| 招聘简历 | 简历经历解析、技能标签映射、岗位/人才画像匹配。 |
-| **`marketing`** | 市场营销 | 用户旅程地图 (CJM)、品牌定位分析、营销活动链路。 |
-| **`supply_chain`**| 物流/供应链 | 供应链上下游追溯、物流节点关系、库存流转图。 |
| **`agriculture`** | 农业与养殖 | 作物生长周期、病虫害防治、畜牧谱系与饲养标准。 |
-| **`art`** | 艺术与博物馆 | 艺术品流传有序图、艺术流派、博物馆馆藏关系。 |
-| **`automotive`** | 汽车工业 | 汽车参数配置、维保手册提取、零部件组成。 |
| **`biology`** | 生物科学 | 生物分类学、蛋白质相互作用、基因调控网络。 |
| **`chemistry`** | 化学科学 | 分子结构、化学反应方程式路径、材料物化特性。 |
| **`physics`** | 物理与天文 | 天体运行系统、量子物理现象、电路逻辑连接图。 |
-| **`construction`**| 建筑工程 | 建筑信息模型 (BIM) 实体、施工进度节点、材料明细。 |
-| **`cybersecurity`**| 网络安全 | 攻击路径图、威胁情报 STIX 模型、系统漏洞分析。 |
| **`environment`** | 环境与气候 | 气象变化趋势、碳排放追踪、生态系统食物网。 |
-| **`fashion`** | 时尚设计 | 时装系列设计元素、流行趋势周期、单品构成。 |
| **`movie`** | 影视 | 剧本节拍器拆解、演员与幕后人员网络、影评情感。 |
-| **`music`** | 音乐 | 乐理结构、专辑曲目分类、流派演进。 |
| **`news`** | 新闻传媒 | 事件因果链、新闻专题聚类、深度报道事实板。 |
-| **`politics`** | 政治与政府 | 政策影响分析、地缘政治关系网、选举数据地图。 |
-| **`real_estate`** | 房地产 | 房源户型特征、租赁协议要素、土地规划关系。 |
+| **`tcm`** | **中医 (特色)** | 中草药方剂、经络穴位、辨证施治过程等。 |
+| **`wuxia`** | **武侠/修真 (特色)** | 门派恩怨、功法境界、修仙等级体系、法宝神兵。 |
-*(注:英文版对应的 31 个通用领域在中文版中均有映射文件夹,中医与武侠为中文版独有)*
+*(注:英文版对应的 16 个通用领域在中文版中均有映射文件夹,中医与武侠为中文版独有)*
## 快速上手
```python
diff --git a/hyperextract/templates/zh/__init__.py b/hyperextract/templates/zh/__init__.py
index ada7ed1..db5ed8b 100644
--- a/hyperextract/templates/zh/__init__.py
+++ b/hyperextract/templates/zh/__init__.py
@@ -2,20 +2,10 @@ from .general.knowledge_graph import KnowledgeGraph
from .general.concept_map import ConceptMap
from .general.logic_graph import LogicGraph
from .agriculture.crop import CropStatusGraph
-from .agriculture.supply_chain import AgriSupplyChainGraph
from .agriculture.soil import SoilNutritionGraph
-from .art.history import ArtHistoryGraph
-from .art.style import ArtStyleGraph
-from .art.exhibition import ExhibitionGraph
-from .automotive.vehicle import VehicleSpecGraph
-from .automotive.diagnosis import VehicleDiagnosisGraph
-from .automotive.market import AutoMarketGraph
from .biology.taxonomy import TaxonomyGraph
from .biology.network import BiologicalNetwork
from .biology.metabolic import MetabolicHypergraph
-from .business.equity import EquityStructureGraph
-from .business.strategy import StrategicChainGraph
-from .business.ecosystem import MarketEcosystemHyper
from .chemistry.molecular import MolecularStructureGraph
from .chemistry.reaction import ChemicalReactionHyper
from .chemistry.protocol import LabProtocolTemporal
@@ -25,20 +15,10 @@ __all__ = [
"ConceptMap",
"LogicGraph",
"CropStatusGraph",
- "AgriSupplyChainGraph",
"SoilNutritionGraph",
- "ArtHistoryGraph",
- "ArtStyleGraph",
- "ExhibitionGraph",
- "VehicleSpecGraph",
- "VehicleDiagnosisGraph",
- "AutoMarketGraph",
"TaxonomyGraph",
"BiologicalNetwork",
"MetabolicHypergraph",
- "EquityStructureGraph",
- "StrategicChainGraph",
- "MarketEcosystemHyper",
"MolecularStructureGraph",
"ChemicalReactionHyper",
"LabProtocolTemporal",
diff --git a/hyperextract/templates/zh/art/__init__.py b/hyperextract/templates/zh/art/__init__.py
deleted file mode 100644
index 96cb725..0000000
--- a/hyperextract/templates/zh/art/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .provenance import ProvenanceGraph
-from .movement import ArtMovementGraph
-from .exhibition import ExhibitionGraph
-
-__all__ = [
- "ProvenanceGraph",
- "ArtMovementGraph",
- "ExhibitionGraph",
-]
diff --git a/hyperextract/templates/zh/art/exhibition.py b/hyperextract/templates/zh/art/exhibition.py
deleted file mode 100644
index 6000d8c..0000000
--- a/hyperextract/templates/zh/art/exhibition.py
+++ /dev/null
@@ -1,152 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. 模式定义 (Schema)
-# ==============================================================================
-
-
-class ExhibitionEntity(BaseModel):
- """艺术展览的组成要素(策展人、主题、作品、场馆、赞助商)。"""
-
- name: str = Field(description="展览名称、艺术家姓名或策展人姓名。")
- type: str = Field(
- description="类型:'展览'、'主题'、'艺术品'、'策展人'、'场馆'、'合作伙伴'。"
- )
- details: Optional[str] = Field(description="展览日期、理念说明或地理位置。")
-
-
-class ExhibitionRelation(BaseModel):
- """展览策划与执行中的连接关系。"""
-
- source: str = Field(description="组织或控制实体。")
- target: str = Field(description="相关的作品 or 参与者。")
- relation_type: str = Field(
- description="类型:'策展自'、'参展于'、'举办于'、'赞助自'、'主题关联至'。"
- )
- context: Optional[str] = Field(description="具体展厅、借展条件或主题子组。")
-
-
-# ==============================================================================
-# 2. 提示词 (Prompts)
-# ==============================================================================
-
-EXHIBITION_GRAPH_PROMPT = (
- "你是一位博物馆馆长和展览设计师。请映射艺术展览的结构图谱。\n\n"
- "准则:\n"
- "- 识别展览名称、核心主题以及参展的艺术家和作品。\n"
- "- 将作品连接到其所属的主题板块以及负责的策展人。\n"
- "- 捕捉场馆和赞助商等物流信息。"
-)
-
-EXHIBITION_NODE_PROMPT = (
- "提取展览组件:展览名称、场馆、主题、具体艺术品以及关键人员(策展人、设计师)。"
-)
-
-EXHIBITION_EDGE_PROMPT = (
- "连接展览内的各项内容。展示哪些作品属于哪个主题,并识别举办场馆和赞助商。"
-)
-
-# ==============================================================================
-# 3. 模板类
-# ==============================================================================
-
-
-class ExhibitionGraph(AutoGraph[ExhibitionEntity, ExhibitionRelation]):
- """
- 展览策划图谱模板,用于策划筹备和展览分析。
-
- 追踪艺术品、主题、策展人和场馆之间的关系。
-
- 示例:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = ExhibitionGraph(llm_client=llm, embedder=embedder)
- >>> text = "'超现实之梦'展览由 Alice Smith 在泰特现代美术馆策划。"
- >>> graph.feed_text(text)
- >>> print(graph.edges) # 展览关联信息
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化展览策划图谱模板。
-
- Args:
- llm_client: 语言模型客户端。
- embedder: 嵌入模型。
- extraction_mode: 提取模式,"one_stage" 或 "two_stage"。
- chunk_size: 分块大小。
- chunk_overlap: 分块重叠大小。
- max_workers: 最大并行工作线程数。
- verbose: 是否开启详细日志。
- **kwargs: 传递给 AutoGraph 的额外参数。
- """
- super().__init__(
- node_schema=ExhibitionEntity,
- edge_schema=ExhibitionRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.strip()}"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=EXHIBITION_GRAPH_PROMPT,
- prompt_for_node_extraction=EXHIBITION_NODE_PROMPT,
- prompt_for_edge_extraction=EXHIBITION_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: ExhibitionEntity) -> str:
- info = f" ({node.type})" if getattr(node, "type", None) else ""
- return f"{node.name}{info}"
-
- def edge_label_extractor(edge: ExhibitionRelation) -> str:
- return f"{edge.relation_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/art/movement.py b/hyperextract/templates/zh/art/movement.py
deleted file mode 100644
index 8def0cd..0000000
--- a/hyperextract/templates/zh/art/movement.py
+++ /dev/null
@@ -1,150 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. 模式定义 (Schema)
-# ==============================================================================
-
-
-class MovementEntity(BaseModel):
- """艺术流派、风格、艺术家或具有影响力的作品。"""
-
- name: str = Field(description="流派名称(如:印象派)或艺术家姓名。")
- category: str = Field(
- description="类别:'流派'、'风格'、'艺术家'、'宣言'、'代表作'。"
- )
- attributes: Optional[str] = Field(description="核心特征、常用媒介或流行时期。")
-
-
-class ArtInfluenceEdge(BaseModel):
- """艺术史中的影响、衍生或成员关系。"""
-
- source: str = Field(description="前驱流派 or 影响源。")
- target: str = Field(description="后继流派、具体作品或艺术家。")
- relation: str = Field(
- description="关系:'影响了'、'创始人'、'成员'、'是对...的反叛'、'衍生自'。"
- )
- details: Optional[str] = Field(description="具体的风格技巧传承或历史事件。")
-
-
-# ==============================================================================
-# 2. 提示词 (Prompts)
-# ==============================================================================
-
-MOVEMENT_GRAPH_PROMPT = (
- "你是一位专门研究艺术风格演变的艺术史学家。请提取艺术流派及其相互影响的图谱。\n\n"
- "准则:\n"
- "- 识别不同的艺术流派以及与之相关的艺术家。\n"
- "- 映射流派之间的演进关系(例如:后印象派对印象派的反思与继承)。\n"
- "- 捕捉宣言或作品中定义的各种核心哲学和视觉特征。"
-)
-
-MOVEMENT_NODE_PROMPT = (
- "提取艺术流派、风格、相关艺术家及关键作品。描述其视觉语言和核心理论。"
-)
-
-MOVEMENT_EDGE_PROMPT = "定义流派与个人之间的联系。使用诸如'是对...的反叛'或'影响了'等关系词。记录代际之间传递的具体技法。"
-
-# ==============================================================================
-# 3. 模板类
-# ==============================================================================
-
-
-class ArtMovementGraph(AutoGraph[MovementEntity, ArtInfluenceEdge]):
- """
- 艺术流派图谱模板,用于映射艺术风格、流派影响力和艺术家所属流派。
-
- 适用于教学应用和艺术史深度分析。
-
- 示例:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = ArtMovementGraph(llm_client=llm, embedder=embedder)
- >>> text = "以莫奈为首的印象派,后来受到了后印象派结构化方法的挑战。"
- >>> graph.feed_text(text)
- >>> print(graph.nodes) # 流派与艺术家详情
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化艺术流派图谱模板。
-
- Args:
- llm_client: 语言模型客户端。
- embedder: 嵌入模型。
- extraction_mode: 提取模式,"one_stage" 或 "two_stage"。
- chunk_size: 分块大小。
- chunk_overlap: 分块重叠大小。
- max_workers: 最大并行工作线程数。
- verbose: 是否开启详细日志。
- **kwargs: 传递给 AutoGraph 的额外参数。
- """
- super().__init__(
- node_schema=MovementEntity,
- edge_schema=ArtInfluenceEdge,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}--({x.relation.lower()})-->{x.target.strip()}"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=MOVEMENT_GRAPH_PROMPT,
- prompt_for_node_extraction=MOVEMENT_NODE_PROMPT,
- prompt_for_edge_extraction=MOVEMENT_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: MovementEntity) -> str:
- info = f" ({node.category})" if getattr(node, "category", None) else ""
- return f"{node.name}{info}"
-
- def edge_label_extractor(edge: ArtInfluenceEdge) -> str:
- return f"{edge.relation}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/art/provenance.py b/hyperextract/templates/zh/art/provenance.py
deleted file mode 100644
index 8380239..0000000
--- a/hyperextract/templates/zh/art/provenance.py
+++ /dev/null
@@ -1,148 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. 模式定义 (Schema)
-# ==============================================================================
-
-
-class ArtEntity(BaseModel):
- """艺术领域实体(如:作品、艺术家、博物馆、私人收藏家、拍卖行)。"""
-
- name: str = Field(description="艺术品名称、人物姓名或机构名称。")
- category: str = Field(
- description="类别:'艺术品'、'艺术家'、'收藏家'、'机构'、'地理位置'。"
- )
- description: Optional[str] = Field(description="材质、创作日期或历史地位描述。")
-
-
-class ProvenanceEdge(BaseModel):
- """艺术品的所有权或流转记录。"""
-
- source: str = Field(description="原持有者或起始地点。")
- target: str = Field(description="新持有者或目的地。")
- transfer_type: str = Field(
- description="流转类型:'收购自'、'出售给'、'捐赠予'、'借展给'、'委托'。"
- )
- specification: Optional[str] = Field(description="流转日期、成交价格或拍卖详情。")
-
-
-# ==============================================================================
-# 2. 提示词 (Prompts)
-# ==============================================================================
-
-PROVENANCE_GRAPH_PROMPT = (
- "你是一位资深的艺术史学家和流传考证专家。请提取艺术品的流传历程图谱。\n\n"
- "准则:\n"
- "- 识别核心艺术品及其所有历史持有者(私人收藏家、博物馆、王室)。\n"
- "- 按照时间顺序映射所有权转移(出售、继承、掠夺、捐赠)。\n"
- "- 捕捉提到的具体日期、成交价格及传承细节。"
-)
-
-PROVENANCE_NODE_PROMPT = "提取艺术实体:识别作品、艺术家、收藏家、博物馆和拍卖行。标注作品的材质及相关人物/机构的角色。"
-
-PROVENANCE_EDGE_PROMPT = "构建传承链条。将前任所有者与现任所有者联系起来,标注流转类型(如'捐赠予')。包含流转的时间段和交易详情。"
-
-# ==============================================================================
-# 3. 模板类
-# ==============================================================================
-
-
-class ProvenanceGraph(AutoGraph[ArtEntity, ProvenanceEdge]):
- """
- 艺术品流传图谱模板,用于追踪艺术品和文物的历史所有权。
-
- 适用于艺术品鉴定、馆藏档案管理和历史研究。
-
- 示例:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI(model="gpt-4o")
- >>> embedder = OpenAIEmbeddings()
- >>> graph = ProvenanceGraph(llm_client=llm, embedder=embedder)
- >>> text = "该画作于1815年由卢浮宫从博尔盖塞收藏中购得。"
- >>> graph.feed_text(text)
- >>> print(graph.edges) # 提取的作品所有权转移记录
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化艺术品流传图谱模板。
-
- Args:
- llm_client: 语言模型客户端。
- embedder: 嵌入模型。
- extraction_mode: 提取模式,"one_stage" 或 "two_stage"。
- chunk_size: 分块大小。
- chunk_overlap: 分块重叠大小。
- max_workers: 最大并行工作线程数。
- verbose: 是否开启详细日志。
- **kwargs: 传递给 AutoGraph 的额外参数。
- """
- super().__init__(
- node_schema=ArtEntity,
- edge_schema=ProvenanceEdge,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}--({x.transfer_type.lower()})-->{x.target.strip()}"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=PROVENANCE_GRAPH_PROMPT,
- prompt_for_node_extraction=PROVENANCE_NODE_PROMPT,
- prompt_for_edge_extraction=PROVENANCE_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
-
- def node_label_extractor(node: ArtEntity) -> str:
- info = f" ({node.category})" if getattr(node, "category", None) else ""
- return f"{node.name}{info}"
-
- def edge_label_extractor(edge: ProvenanceEdge) -> str:
- return f"{edge.transfer_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/automotive/__init__.py b/hyperextract/templates/zh/automotive/__init__.py
deleted file mode 100644
index 0184d1b..0000000
--- a/hyperextract/templates/zh/automotive/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .car_spec import CarSpecGraph
-from .maintenance import CarMaintenanceGraph
-from .manufacturing import AutomotiveMfgGraph
-
-__all__ = [
- "CarSpecGraph",
- "CarMaintenanceGraph",
- "AutomotiveMfgGraph",
-]
diff --git a/hyperextract/templates/zh/automotive/car_spec.py b/hyperextract/templates/zh/automotive/car_spec.py
deleted file mode 100644
index 9666886..0000000
--- a/hyperextract/templates/zh/automotive/car_spec.py
+++ /dev/null
@@ -1,141 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. 模式定义 (Schema)
-# ==============================================================================
-
-class CarComponent(BaseModel):
- """车辆组件或参数类别(发动机、电池、自动驾驶、传动系统)。"""
- name: str = Field(description="车型名称或具体组件名称。")
- category: str = Field(
- description="类别:'车型'、'动力系统'、'底盘'、'安全系统'、'信息娱乐'。"
- )
- specification: Optional[str] = Field(description="详细参数(如:马力、电量、版本)。")
-
-class CarSystemRelation(BaseModel):
- """汽车系统间的层级或技术关系。"""
- source: str = Field(description="父系统或车辆型号。")
- target: str = Field(description="子组件或功能。")
- relation_type: str = Field(
- description="类型:'配备'、'动力源自'、'集成'、'兼容'。"
- )
- performance_note: Optional[str] = Field(description="性能指标或集成细节。")
-
-# ==============================================================================
-# 2. 提示词 (Prompts)
-# ==============================================================================
-
-CAR_SPEC_GRAPH_PROMPT = (
- "你是一位汽车工程师和技术评论员。请提取车辆的技术参数图谱。\n\n"
- "准则:\n"
- "- 识别车型及其核心系统(发动机、电机、电池、支驾系统)。\n"
- "- 映射组件层级关系(例如:电池包含电芯,发动机使用涡轮增压器)。\n"
- "- 捕捉可量化的参数,如 0-100km/h 加速、续航里程、峰值功率和扭矩。"
-)
-
-CAR_SPEC_NODE_PROMPT = (
- "提取车型和技术组件。重点关注名称及其可测量的参数指标(如:500马力、80kWh)。"
-)
-
-CAR_SPEC_EDGE_PROMPT = (
- "定义技术依赖关系。展示哪些组件属于哪个汽车系统。使用诸如'动力源自'或'配备'等关系。"
-)
-
-# ==============================================================================
-# 3. 模板类
-# ==============================================================================
-
-class CarSpecGraph(AutoGraph[CarComponent, CarSystemRelation]):
- """
- 汽车参数图谱模板,用于提取车辆技术数据和参数层级。
-
- 适用于竞品分析、汽车垂直媒体和技术文档处理。
-
- 示例:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = CarSpecGraph(llm_client=llm, embedder=embedder)
- >>> text = "特斯拉 Model 3 高性能版采用双电机全轮驱动,可输出 510 马力。"
- >>> graph.feed_text(text)
- >>> print(graph.nodes) # 车辆及其组件参数
- """
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any
- ):
- """
- 初始化汽车参数图谱模板。
-
- Args:
- llm_client: 语言模型客户端。
- embedder: 嵌入模型。
- extraction_mode: 提取模式,"one_stage" 或 "two_stage"。
- chunk_size: 分块大小。
- chunk_overlap: 分块重叠大小。
- max_workers: 最大并行工作线程数。
- verbose: 是否开启详细日志。
- **kwargs: 传递给 AutoGraph 的额外参数。
- """
- super().__init__(
- node_schema=CarComponent,
- edge_schema=CarSystemRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.strip()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CAR_SPEC_GRAPH_PROMPT,
- prompt_for_node_extraction=CAR_SPEC_NODE_PROMPT,
- prompt_for_edge_extraction=CAR_SPEC_EDGE_PROMPT,
- **kwargs
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: CarComponent) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.name }{info}"
-
- def edge_label_extractor(edge: CarSystemRelation) -> str:
- return f"{ edge.relation_type }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/automotive/maintenance.py b/hyperextract/templates/zh/automotive/maintenance.py
deleted file mode 100644
index e09e22d..0000000
--- a/hyperextract/templates/zh/automotive/maintenance.py
+++ /dev/null
@@ -1,137 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. 模式定义 (Schema)
-# ==============================================================================
-
-class MaintenanceItem(BaseModel):
- """保养项目或车辆零部件(机油、刹车片、正时皮带)。"""
- item_name: str = Field(description="零部件名称或保养服务名称。")
- category: str = Field(description="类别:'油液'、'易损件'、'检查项'、'软件更新'。")
- interval: Optional[str] = Field(description="建议的里程或时间间隔(如:5,000公里)。")
-
-class MaintenanceSequence(BaseModel):
- """保养行动的依赖或序列关系。"""
- source: str = Field(description="触发保养的部件或初始状态。")
- target: str = Field(description="保养任务或更换的零部件。")
- action: str = Field(description="动作:'更换'、'检查'、'清洗'、'校准'。")
- precaution: Optional[str] = Field(description="安全警告或所需工具。")
-
-# ==============================================================================
-# 2. 提示词 (Prompts)
-# ==============================================================================
-
-MAINTENANCE_GRAPH_PROMPT = (
- "你是一位资深技师和售后顾问。请提取车辆保养与维修图谱。\n\n"
- "准则:\n"
- "- 识别手册或日志中提到的具体保养项目和零部件。\n"
- "- 映射触发保养的条件(如:里程、报警灯、视觉检查结果)。\n"
- "- 将零部件连接到对应的系统及所需的维保动作。"
-)
-
-MAINTENANCE_NODE_PROMPT = (
- "提取保养任务、零部件和油液类型。如果指定了建议间隔,请一并记录。"
-)
-
-MAINTENANCE_EDGE_PROMPT = (
- "将触发条件与动作联系起来。展示哪些部件需要更换或检查。记录每项任务的具体注意事项。"
-)
-
-# ==============================================================================
-# 3. 模板类
-# ==============================================================================
-
-class CarMaintenanceGraph(AutoGraph[MaintenanceItem, MaintenanceSequence]):
- """
- 汽车保养图谱模板,用于处理车辆服务手册、维修指南和保养日志。
-
- 有助于整理维保计划和零部件兼容性信息。
-
- 示例:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = CarMaintenanceGraph(llm_client=llm, embedder=embedder)
- >>> text = "每10,000公里更换发动机机油和机滤,使用 0W-20 全合成机油。"
- >>> graph.feed_text(text)
- >>> print(graph.nodes) # 保养项目详情
- """
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any
- ):
- """
- 初始化汽车保养图谱模板。
-
- Args:
- llm_client: 语言模型客户端。
- embedder: 嵌入模型。
- extraction_mode: 提取模式,"one_stage" 或 "two_stage"。
- chunk_size: 分块大小。
- chunk_overlap: 分块重叠大小。
- max_workers: 最大并行工作线程数。
- verbose: 是否开启详细日志。
- **kwargs: 传递给 AutoGraph 的额外参数。
- """
- super().__init__(
- node_schema=MaintenanceItem,
- edge_schema=MaintenanceSequence,
- node_key_extractor=lambda x: x.item_name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.action.lower()})-->{x.target.strip()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=MAINTENANCE_GRAPH_PROMPT,
- prompt_for_node_extraction=MAINTENANCE_NODE_PROMPT,
- prompt_for_edge_extraction=MAINTENANCE_EDGE_PROMPT,
- **kwargs
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: MaintenanceItem) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.item_name }{info}"
-
- def edge_label_extractor(edge: MaintenanceSequence) -> str:
- return f"{ edge.action }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/automotive/manufacturing.py b/hyperextract/templates/zh/automotive/manufacturing.py
deleted file mode 100644
index 29afdb7..0000000
--- a/hyperextract/templates/zh/automotive/manufacturing.py
+++ /dev/null
@@ -1,141 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. 模式定义 (Schema)
-# ==============================================================================
-
-class ManufacturingEntity(BaseModel):
- """汽车供应链或生产线上的实体(供应商、工厂、组装环节)。"""
- name: str = Field(description="公司名称、工厂名称或生产工艺名称。")
- category: str = Field(
- description="类别:'供应商'、'工厂'、'工艺'、'原材料'、'质检'。"
- )
- specification: Optional[str] = Field(description="产能、资质认证或交付周期。")
-
-class SupplyChainRelation(BaseModel):
- """汽车制造中参与方与环节之间的连接关系。"""
- source: str = Field(description="上游供应商或工艺环节。")
- target: str = Field(description="下游工厂或产品。")
- relation_type: str = Field(
- description="类型:'供货给'、'组装于'、'处理自'、'认证用于'。"
- )
- logistics_details: Optional[str] = Field(description="贸易术语、运输时间或供应量承诺。")
-
-# ==============================================================================
-# 2. 提示词 (Prompts)
-# ==============================================================================
-
-MFG_GRAPH_PROMPT = (
- "你是一位大型汽车制造商的供应链分析师。请提取汽车制造与供应链图谱。\n\n"
- "准则:\n"
- "- 识别各级供应商(一级、二级供应商)及其提供的零部件。\n"
- "- 映射从原材料到整车下线的组装序列。\n"
- "- 捕捉工厂的地理位置及特定的物流限制条件。"
-)
-
-MFG_NODE_PROMPT = (
- "提取供应商、制造工厂和关键生产步骤。标注它们在车辆生产周期中的角色。"
-)
-
-MFG_EDGE_PROMPT = (
- "定义物料流和工艺流。连接供应商到组装工厂,以及工厂到最终的整车交付环节。"
-)
-
-# ==============================================================================
-# 3. 模板类
-# ==============================================================================
-
-class AutomotiveMfgGraph(AutoGraph[ManufacturingEntity, SupplyChainRelation]):
- """
- 汽车制造图谱模板,用于生产线流程、供应链物流和供应商管理。
-
- 有助于可视化全球汽车制造生态系统中的依赖关系。
-
- 示例:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> llm = ChatOpenAI()
- >>> embedder = OpenAIEmbeddings()
- >>> graph = AutomotiveMfgGraph(llm_client=llm, embedder=embedder)
- >>> text = "宁德时代向特斯拉柏林超级工厂供应电池芯,用于 Model Y 的组装。"
- >>> graph.feed_text(text)
- >>> print(graph.edges) # 供应链流动信息
- """
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any
- ):
- """
- 初始化汽车制造图谱模板。
-
- Args:
- llm_client: 语言模型客户端。
- embedder: 嵌入模型。
- extraction_mode: 提取模式,"one_stage" 或 "two_stage"。
- chunk_size: 分块大小。
- chunk_overlap: 分块重叠大小。
- max_workers: 最大并行工作线程数。
- verbose: 是否开启详细日志。
- **kwargs: 传递给 AutoGraph 的额外参数。
- """
- super().__init__(
- node_schema=ManufacturingEntity,
- edge_schema=SupplyChainRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.strip()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=MFG_GRAPH_PROMPT,
- prompt_for_node_extraction=MFG_NODE_PROMPT,
- prompt_for_edge_extraction=MFG_EDGE_PROMPT,
- **kwargs
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: ManufacturingEntity) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.name }{info}"
-
- def edge_label_extractor(edge: SupplyChainRelation) -> str:
- return f"{ edge.relation_type }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/business/__init__.py b/hyperextract/templates/zh/business/__init__.py
deleted file mode 100644
index 8da251c..0000000
--- a/hyperextract/templates/zh/business/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .equity import EquityStructureGraph
-from .strategy import StrategicChainGraph
-from .ecosystem import MarketEcosystemHyper
-
-__all__ = [
- "EquityStructureGraph",
- "StrategicChainGraph",
- "MarketEcosystemHyper",
-]
diff --git a/hyperextract/templates/zh/business/ecosystem.py b/hyperextract/templates/zh/business/ecosystem.py
deleted file mode 100644
index 8ec4a84..0000000
--- a/hyperextract/templates/zh/business/ecosystem.py
+++ /dev/null
@@ -1,139 +0,0 @@
-from typing import List, Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoHypergraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class IndustryParticipant(BaseModel):
- """
- 产业生态系统中的实体(公司、监管机构、非政府组织)。
- """
-
- name: str = Field(description="参与者名称(如:'台积电'、'欧盟委员会')。")
- rank: Optional[str] = Field(
- None, description="市场地位:'一级供应商'、'市场领导者'、'挑战者'。"
- )
- technological_focus: Optional[str] = Field(
- None, description="主要技术或领域(如:'5G'、'光刻技术')。"
- )
-
-
-class CollectiveAlliance(BaseModel):
- """
- 代表财团、联盟、市场卡特尔或联合标准制定机构的超边。
- """
-
- alliance_id: str = Field(description="联盟的名称或标识符(如:'芯片联盟')。")
- participants: List[str] = Field(
- description="在该团体中进行合作或竞争的所有实体的名称列表。"
- )
- shared_goal: str = Field(description="目标:'标准化'、'反垄断审查'、'联合研发'。")
- governance: Optional[str] = Field(None, description="参与原则或领导实体。")
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-ECOSYSTEM_CONSOLIDATED_PROMPT = (
- "你是一位市场战略家和反垄断专家。请提取市场生态系统的动态。\n\n"
- "规则:\n"
- "- 使用超边(Hyperedges)将参与合资企业或联盟的多个实体(3个或更多)联系起来。\n"
- "- 捕获非二元的群体关系,如卡特尔、产业集群和监管联盟。\n"
- "- 清晰地定义这些群体的共同目标。"
-)
-
-ECOSYSTEM_NODE_PROMPT = (
- "你是一位市场分析师。你的任务是识别产业生态系统中的所有参与者(节点)。\n\n"
- "提取规则:\n"
- "- 识别提及的公司、非政府组织(NGO)和监管机构。\n"
- "- 捕捉它们的市场地位(Rank)和主要技术关注领域。\n"
- "- 侧重于识别那些以群体形式进行合作或竞争的角色。\n"
- "- 在此阶段请勿识别联盟、合资企业或财团。"
-)
-
-ECOSYSTEM_EDGE_PROMPT = (
- "你是一位市场分析师。根据产业参与者清单,提取集体关系(超边)。\n\n"
- "提取规则:\n"
- "- 将多个参与者(3个或更多)归入联盟、标准制定机构或合资企业。\n"
- "- 为整个群体定义共同的目标(如:'研发'、'反垄断审查')。\n"
- "- 记录提及的任何治理结构或领导参与者。\n"
- "- 确保群体的每个成员都选自提供的参与者清单。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class MarketEcosystemHyper(AutoHypergraph[IndustryParticipant, CollectiveAlliance]):
- """
- 用于产业集群、生态系统联盟和复杂市场协同的超图模板。
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- super().__init__(
- node_schema=IndustryParticipant,
- edge_schema=CollectiveAlliance,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: x.alliance_id.strip(),
- nodes_in_edge_extractor=lambda x: tuple(x.participants),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=ECOSYSTEM_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=ECOSYSTEM_NODE_PROMPT,
- prompt_for_edge_extraction=ECOSYSTEM_EDGE_PROMPT,
- **kwargs,
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: IndustryParticipant) -> str:
- return f"{ node.name }"
-
- def edge_label_extractor(edge: CollectiveAlliance) -> str:
- return f"{ edge.alliance_id }"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/business/equity.py b/hyperextract/templates/zh/business/equity.py
deleted file mode 100644
index b55d6c8..0000000
--- a/hyperextract/templates/zh/business/equity.py
+++ /dev/null
@@ -1,148 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class StakeholderNode(BaseModel):
- """
- 在业务中持有股份或拥有控制权的法人或自然人。
- """
-
- name: str = Field(description="公司注册全称或个人全名。")
- type: str = Field(
- description="类型:'自然人'、'上市公司'、'私有公司'、'国有企业'、'风险投资'。"
- )
- jurisdiction: Optional[str] = Field(
- None, description="注册地或居住地所属的国家或地区。"
- )
- is_ubo: bool = Field(False, description="该实体是否被识别为最终受益人(UBO)。")
-
-
-class OwnershipRelation(BaseModel):
- """
- 两个实体之间的持股或控制联系。
- """
-
- source: str = Field(description="股东或母公司实体。")
- target: str = Field(description="被拥有或控制的公司(子公司)。")
- stake_percentage: Optional[float] = Field(
- None, description="持股比例(0.0 到 100.0)。"
- )
- control_type: str = Field(
- "Equity", description="联系性质:'股权收购'、'投票权'、'代持股份'、'直接控制'。"
- )
- as_of_date: Optional[str] = Field(
- None, description="记录该所有权的时间点或财务周期。"
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-EQUITY_CONSOLIDATED_PROMPT = (
- "你是一位企业法务助理和财务分析师。请提取复杂的股权结构。\n\n"
- "规则:\n"
- "- 优先识别每个实体持有的股份比例。\n"
- "- 区分直接股东和最终受益人(UBO)。\n"
- "- 映射从母公司到最小子公司的层级结构。"
-)
-
-EQUITY_NODE_PROMPT = (
- "你是一位企业法务助理和财务分析师。你的任务是识别出公司结构中的所有利益相关者(节点)。\n\n"
- "提取规则:\n"
- "- 识别公司、投资基金、政府机构和自然人个人。\n"
- "- 将其分类为 '自然人'、'上市公司'、'私有公司' 等。\n"
- "- 在明确的情况下,标记被识别为最终受益人(UBO)的实体。\n"
- "- 在此阶段请勿提取持股比例或控制联系。"
-)
-
-EQUITY_EDGE_PROMPT = (
- "你是一位企业法务助理和财务分析师。根据利益相关者清单,映射所有权和控制关系(边)。\n\n"
- "提取规则:\n"
- "- 提取股东(源)与公司(目标)之间持有的确切股份比例。\n"
- "- 指定控制的性质(如:投票权、直接控制、代持股份)。\n"
- "- 仅在提供的利益相关者清单中的成员之间建立边。\n"
- "- 捕获所有权记录的 '截至日期' 或历史时期。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class EquityStructureGraph(AutoGraph[StakeholderNode, OwnershipRelation]):
- """
- 用于企业登记、并购分析及 KYC/UBO 追踪的高精度模板。
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- super().__init__(
- node_schema=StakeholderNode,
- edge_schema=OwnershipRelation,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}->{x.target.strip()}_at_{x.stake_percentage}%"
- ),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=EQUITY_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=EQUITY_NODE_PROMPT,
- prompt_for_edge_extraction=EQUITY_EDGE_PROMPT,
- **kwargs,
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: StakeholderNode) -> str:
- info = f" ({ node.type })" if getattr(node, "type", None) else ""
- return f"{ node.name }{info}"
-
- def edge_label_extractor(edge: OwnershipRelation) -> str:
- return f"{edge.control_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/business/strategy.py b/hyperextract/templates/zh/business/strategy.py
deleted file mode 100644
index f304ec0..0000000
--- a/hyperextract/templates/zh/business/strategy.py
+++ /dev/null
@@ -1,146 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoTemporalGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class StrategicNode(BaseModel):
- """
- 组织实体、业务部门或战略目标。
- """
-
- name: str = Field(description="战略单元或目标的名称(如:'Azure'、'碳中和')。")
- scope: str = Field(description="范围:'全球'、'区域'、'特定产品'、'集团层面'。")
- importance: Optional[str] = Field(
- None, description="重要性:'核心业务'、'新兴市场'、'传统存量'。"
- )
-
-
-class StrategicPivot(BaseModel):
- """
- 代表战略重点转移、投资或剥离的时间性转变。
- """
-
- source: str = Field(description="原始关注领域或施加动作的实体。")
- target: str = Field(description="新的目标、方向或目的。")
- action: str = Field(
- description="战略动作:'转型为'、'收购'、'剥离'、'合并'、'关停'、'规模化'。"
- )
- timestamp: str = Field(
- description="转型发生的时间(如:'2022年第四季度'、'2025-01-01')。"
- )
- rationale: Optional[str] = Field(
- None, description="举措背后的原因(市场压力、CEO 变动、技术变革)。"
- )
- budget: Optional[str] = Field(None, description="与转型相关的财务承诺或预算。")
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-STRATEGY_CONSOLIDATED_PROMPT = (
- "你是一位管理顾问和行业分析师。请提取公司战略的演变过程。\n\n"
- "规则:\n"
- "- 专注于 '转型(Pivots)':从一个业务领域向另一个业务领域的转移。\n"
- "- 将特定动作与具体时间戳联系起来。\n"
- "- 捕获每次战略变革背后的深层原因(rationale)。"
-)
-
-STRATEGY_NODE_PROMPT = (
- "你是一位管理顾问。你的任务是识别战略实体和目标(节点)。\n\n"
- "提取规则:\n"
- "- 识别业务部门、特定产品以及公司高层目标(如:'可持续发展')。\n"
- "- 确定每个单元的范围和组织重要性。\n"
- "- 专注于代表战略重点领域的名词和名词短语。\n"
- "- 在此阶段请勿提取战略性动作或转型。"
-)
-
-STRATEGY_EDGE_PROMPT = (
- "你是一位管理顾问。根据已识别的战略单元,映射战略随时间推移的转变(边)。\n\n"
- "提取规则:\n"
- "- 提取 '转型 (Pivots)' 以及如 '收购'、'剥离' 或 '关停' 之类的动作。\n"
- "- 将源(旧重点)与目标(新重点)通过精确的时间戳联系起来。\n"
- "- 捕捉深层原因以及提及的任何财务承诺(预算)。\n"
- "- 仅引用战略单元清单中存在的节点。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class StrategicChainGraph(AutoTemporalGraph[StrategicNode, StrategicPivot]):
- """
- 用于追踪公司战略、长期组织变革和市场表现的时序模板。
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- super().__init__(
- node_schema=StrategicNode,
- edge_schema=StrategicPivot,
- node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip()}-({x.action})->{x.target.strip()}"
- ),
- time_in_edge_extractor=lambda x: x.timestamp.strip(),
- nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=STRATEGY_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=STRATEGY_NODE_PROMPT,
- prompt_for_edge_extraction=STRATEGY_EDGE_PROMPT,
- **kwargs,
- )
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: StrategicNode) -> str:
- return f"{node.name}"
-
- def edge_label_extractor(edge: StrategicPivot) -> str:
- return f"{edge.action}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/construction/__init__.py b/hyperextract/templates/zh/construction/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/construction/project_timeline.py b/hyperextract/templates/zh/construction/project_timeline.py
deleted file mode 100644
index d07f2b1..0000000
--- a/hyperextract/templates/zh/construction/project_timeline.py
+++ /dev/null
@@ -1,160 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoTemporalGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class MilestoneTaskNode(BaseModel):
- """
- 建筑项目进度计划中的特定任务、里程碑或阶段。
- """
- task_name: str = Field(description="任务操作名称(例如:'基础浇筑')。")
- phase: str = Field(description="项目阶段:'前期准备'、'下部结构'、'主体结构'、'装修装饰'、'竣工验收'。")
- duration_estimate: Optional[str] = Field(None, description="预估时间(例如:'14天'、'2个月')。")
- responsible_party: Optional[str] = Field(None, description="负责的分包商或团队负责人。")
-
-class TemporalDependency(BaseModel):
- """
- 表示任务权序的基于时间的连接。
- """
- source: str = Field(description="前置任务。")
- target: str = Field(description="后置任务。")
- timestamp: str = Field(
- description="时间标记或步骤编号(例如:'阶段1'、'步骤2'或具体日期/时间)。"
- )
- dependency_type: str = Field(
- "完成到开始",
- description="类型:'完成到开始 (FS)'、'开始到开始 (SS)'、'完成到完成 (FF)'、'开始到完成 (SF)'。"
- )
- lag_time: Optional[str] = Field(None, description="任务间的等待时间(例如:'混凝土养护3天')。")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CONSTRUCTION_TEMPORAL_CONSOLIDATED_PROMPT = (
- "你是一位资深项目管理计划员和计划工程师。请从文本中提取建筑项目的时间进度表。\n\n"
- "提取规则:\n"
- "- 识别细化的任务和重大里程碑作为节点。\n"
- "- 使用行业标准逻辑(FS, SS等)建立清晰的时间依赖关系(边)。\n"
- "- 捕获提到的持续时间和养护/滞后时间。\n"
- "- 重点关注关键路径和测序逻辑。"
-)
-
-CONSTRUCTION_TEMPORAL_NODE_PROMPT = (
- "识别所有任务、计划活动和项目里程碑。\n\n"
- "提取规则:\n"
- "- 提取以动词开头的任务名称。\n"
- "- 将任务归类到项目阶段并识别负责方。\n"
- "- 在此阶段不要识别顺序或依赖关系。"
-)
-
-CONSTRUCTION_TEMPORAL_EDGE_PROMPT = (
- "根据任务在时间上的逻辑顺序进行连接。\n\n"
- "提取规则:\n"
- "- 为每个链接定义前导任务和后续任务。\n"
- "- 分类链接类型(如果未指定,默认为FS)。\n"
- "- 识别任何要求的滞后时间(例如:等待检验或干燥)。\n"
- "- 仅在已识别的列表任务之间创建连接。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class ProjectSchedulingTimeline(AutoTemporalGraph[MilestoneTaskNode, TemporalDependency]):
- """
- 用于提取项目计划和关键路径法(CPM)网络的模板。
-
- 适用于进度风险分析、延期跟踪和进度监控。
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> timeline = ProjectSchedulingTimeline(llm_client=llm, embedder=embedder)
- >>> timeline.feed_text("土方开挖必须在浇筑基础底板之前完成。")
- >>> print(timeline.nodes) # 获取任务和里程碑
- >>> print(timeline.edges) # 获取时间依赖关系
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化项目进度时间轴模板。
-
- Args:
- llm_client (BaseChatModel): 用于进度提取的大模型客户端。
- embedder (Embeddings): 用于任务和依赖项去重的嵌入模型。
- extraction_mode (str, optional): 'one_stage' 为联合提取,
- 'two_stage' 为分步提取。默认为 "one_stage"。
- chunk_size (int, optional): 单个文本块的最大字符数。默认为 2048。
- chunk_overlap (int, optional): 文本块之间的重叠字符数。默认为 256。
- max_workers (int, optional): 并行处理的工作线程数。默认为 10。
- verbose (bool, optional): 是否启用详细日记记录。默认为 False。
- **kwargs (Any): 传递给 AutoTemporalGraph 基类的其他参数。
- """
- super().__init__(
- node_schema=MilestoneTaskNode,
- edge_schema=TemporalDependency,
- node_key_extractor=lambda x: x.task_name.strip().lower(),
- edge_key_extractor=lambda x: f"{x.source.strip().lower()}--[{x.dependency_type}]-->{x.target.strip().lower()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip().lower(), x.target.strip().lower()),
- time_in_edge_extractor=lambda x: x.timestamp.strip(),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CONSTRUCTION_TEMPORAL_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CONSTRUCTION_TEMPORAL_NODE_PROMPT,
- prompt_for_edge_extraction=CONSTRUCTION_TEMPORAL_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: MilestoneTaskNode) -> str:
- info = f" ({node.phase})" if getattr(node, "phase", None) else ""
- return f"{node.task_name}{info}"
-
- def edge_label_extractor(edge: TemporalDependency) -> str:
- return f"{edge.dependency_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/construction/resource_registry.py b/hyperextract/templates/zh/construction/resource_registry.py
deleted file mode 100644
index 9b3d927..0000000
--- a/hyperextract/templates/zh/construction/resource_registry.py
+++ /dev/null
@@ -1,108 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoSet
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class ConstructionResource(BaseModel):
- """
- 分配给建筑项目的特定资产、材料或劳动力类别。
- """
- resource_name: str = Field(description="资源名称(例如:'塔吊 TC-01'、'粉煤灰混凝土')。")
- category: str = Field(description="类别:'重型机械'、'散装材料'、'技工'、'固定设备'、'易耗品'。")
- quantity: Optional[str] = Field(None, description="现场当前数量或容量。")
- supplier_or_team: Optional[str] = Field(None, description="提供或管理此资源的实体。")
- status: str = Field("在用", description="当前运营状态:'在用'、'闲置'、'维护'、'用尽'。")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CONSTRUCTION_RESOURCE_PROMPT = (
- "你是一位建筑物流和采购经理。请从文本中提取所有项目资源和资产。\n\n"
- "提取规则:\n"
- "- 识别机械、材料和劳动力类型。\n"
- "- 捕获提到的具体数量和供应商名称。\n"
- "- 自动合并相似项(例如:'挖掘机 A' 和 '那台主要的挖掘机' 如果指代同一个单元,应合并)。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class ResourceRegistrySet(AutoSet[ConstructionResource]):
- """
- 用于管理统一的工地资产和材料清单的模板。
-
- 使用 AutoSet 自动去重并合并多段文本中提到的资源信息。
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> registry = ResourceRegistrySet(llm_client=llm, embedder=embedder)
- >>> registry.feed_text("现场有两台来自利勃海尔的50吨起重机。")
- >>> registry.feed_text("这些利勃海尔起重机目前正在进行维修。")
- >>> print(registry.items) # 信息将合并到单个起重机资源中
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化资源登记表模板。
-
- Args:
- llm_client (BaseChatModel): 用于资源提取的大模型客户端。
- embedder (Embeddings): 用于资源去重的嵌入模型。
- chunk_size (int, optional): 单个文本块的最大字符数。默认为 2048。
- chunk_overlap (int, optional): 文本块之间的重叠字符数。默认为 256。
- max_workers (int, optional): 并行处理的工作线程数。默认为 10。
- verbose (bool, optional): 是否启用详细日记记录。默认为 False。
- **kwargs (Any): 传递给 AutoSet 基类的其他参数。
- """
- super().__init__(
- item_schema=ConstructionResource,
- key_extractor=lambda x: x.resource_name.strip().lower(),
- llm_client=llm_client,
- embedder=embedder,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CONSTRUCTION_RESOURCE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_for_search: int = 3,
- top_k_for_chat: int = 3,
- ) -> None:
- """
- Visualize the collection using OntoSight.
-
- Args:
- top_k_for_search (int): Number of items to retrieve for search context. Default 3.
- top_k_for_chat (int): Number of items to retrieve for chat context. Default 3.
- """
- def item_label_extractor(item: ConstructionResource) -> str:
- info = f" ({ item.category })" if getattr(item, "category", None) else ""
- return f"{ item.resource_name }{info}"
-
- super().show(
- item_label_extractor=item_label_extractor,
- top_k_for_search=top_k_for_search,
- top_k_for_chat=top_k_for_chat,
- )
diff --git a/hyperextract/templates/zh/construction/site_topology.py b/hyperextract/templates/zh/construction/site_topology.py
deleted file mode 100644
index 600edf7..0000000
--- a/hyperextract/templates/zh/construction/site_topology.py
+++ /dev/null
@@ -1,157 +0,0 @@
-from typing import Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoSpatialGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-class PhysicalElementNode(BaseModel):
- """
- 建筑工地内的特定物理构件或区域。
- """
- name: str = Field(description="元素的唯一标识符或名称(例如:'A1柱'、'B区')。")
- element_type: str = Field(description="类型:'结构件'、'机电构件'、'建筑构件'、'临时设施'、'安全区域'。")
- location_coordinates: Optional[str] = Field(None, description="空间坐标或相对轴线位置(例如:'X:10, Y:20, Z:5')。")
- material_properties: Optional[str] = Field(None, description="关键材料信息(例如:'C30混凝土'、'S355钢材')。")
-
-class SpatialRelation(BaseModel):
- """
- 两个物理元素之间的空间或拓扑连接关系。
- """
- source: str = Field(description="参考元素。")
- target: str = Field(description="关联元素。")
- location_marker: str = Field(
- description="该关系所在的物理位置或区域(例如:'3层'、'北翼'、'A-1轴线')。"
- )
- relation_type: str = Field(
- description="类型:'支撑'、'相邻'、'包含于'、'相交'、'对齐'。"
- )
- offset_distance: Optional[float] = Field(None, description="元素间的物理距离(单位:米)。")
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CONSTRUCTION_SPATIAL_CONSOLIDATED_PROMPT = (
- "你是一位资深BIM(建筑信息模型)工程师。请从文本中提取建筑项目的空间拓扑结构。\n\n"
- "提取规则:\n"
- "- 识别固定的物理构件及其属性作为节点。\n"
- "- 映射这些构件之间的结构和空间关系(边)。\n"
- "- 如果文本提到了坐标和距离,请务必捕获。\n"
- "- 重点关注物体是如何物理连接或相对定位的。"
-)
-
-CONSTRUCTION_SPATIAL_NODE_PROMPT = (
- "识别所有物理建筑构件和工地区域。\n\n"
- "提取规则:\n"
- "- 提取明确的结构、机械或建筑构件。\n"
- "- 捕获任何提到的位置、楼层或材料信息。\n"
- "- 在此阶段不要识别连接关系。"
-)
-
-CONSTRUCTION_SPATIAL_EDGE_PROMPT = (
- "映射已识别建筑构件之间的空间和结构关系。\n\n"
- "提取规则:\n"
- "- 确定元素A与元素B的关系(例如:A支撑B,A与B相邻)。\n"
- "- 提取提供的物理偏移或距离。\n"
- "- 仅在已识别的列表元素之间创建连接。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-class SiteTopologyGraph(AutoSpatialGraph[PhysicalElementNode, SpatialRelation]):
- """
- 用于构建建筑工地空间数字孪生的模板。
-
- 适用于结构分析、BIM转图谱以及工地物流规划。
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> graph = SiteTopologyGraph(llm_client=llm, embedder=embedder)
- >>> graph.feed_text("基础底板支撑着四个主要的转角柱。")
- >>> print(graph.nodes) # 获取结构构件
- >>> print(graph.edges) # 获取空间关系
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化工地拓扑图谱模板。
-
- Args:
- llm_client (BaseChatModel): 用于拓扑提取的大模型客户端。
- embedder (Embeddings): 用于元素去重的嵌入模型。
- extraction_mode (str, optional): 'one_stage' 为联合提取,
- 'two_stage' 为分步提取。默认为 "one_stage"。
- chunk_size (int, optional): 单个文本块的最大字符数。默认为 2048。
- chunk_overlap (int, optional): 文本块之间的重叠字符数。默认为 256。
- max_workers (int, optional): 并行处理的工作线程数。默认为 10。
- verbose (bool, optional): 是否启用详细日记记录。默认为 False。
- **kwargs (Any): 传递给 AutoSpatialGraph 基类的其他参数。
- """
- super().__init__(
- node_schema=PhysicalElementNode,
- edge_schema=SpatialRelation,
- node_key_extractor=lambda x: x.name.strip().lower(),
- edge_key_extractor=lambda x: f"{x.source.strip().lower()}--{x.relation_type.lower()}--{x.target.strip().lower()}",
- nodes_in_edge_extractor=lambda x: (x.source.strip().lower(), x.target.strip().lower()),
- location_in_edge_extractor=lambda x: x.location_marker.strip(),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CONSTRUCTION_SPATIAL_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CONSTRUCTION_SPATIAL_NODE_PROMPT,
- prompt_for_edge_extraction=CONSTRUCTION_SPATIAL_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: PhysicalElementNode) -> str:
- return f"{node.name}"
-
- def edge_label_extractor(edge: SpatialRelation) -> str:
- return f"{edge.relation_type}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/cybersecurity/__init__.py b/hyperextract/templates/zh/cybersecurity/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/cybersecurity/intrusion_kill_chain.py b/hyperextract/templates/zh/cybersecurity/intrusion_kill_chain.py
deleted file mode 100644
index 5f2c5db..0000000
--- a/hyperextract/templates/zh/cybersecurity/intrusion_kill_chain.py
+++ /dev/null
@@ -1,175 +0,0 @@
-from typing import Optional, List, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoGraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class AttackStageNode(BaseModel):
- """
- 网络攻击中的特定阶段或行动(符合 MITRE ATT&CK 或 Lockheed Kill Chain)。
- """
-
- stage_name: str = Field(
- description="攻击阶段名称(例如:'初始访问'、'权限维持'、'横向移动')。"
- )
- tactics_used: List[str] = Field(
- default_factory=list,
- description="所使用的具体技术或战术(例如:'鱼叉式网络钓鱼'、'哈希传递')。",
- )
- adversary_tools: List[str] = Field(
- default_factory=list,
- description="该阶段使用的恶意软件或工具(例如:'Cobalt Strike'、'Mimikatz')。",
- )
- timestamp: Optional[str] = Field(None, description="行动的时间点或顺序标记。")
-
-
-class KillChainLink(BaseModel):
- """
- 表示攻击从一个阶段向另一个阶段演进的有向链接。
- """
-
- source: str = Field(description="前置攻击阶段。")
- target: str = Field(description="后续攻击阶段。")
- causality: str = Field(
- "演进",
- description="类型:'演进' (标准)、'触发' (自动执行)、'跳转' (基础设施切换)。",
- )
- observables: List[str] = Field(
- default_factory=list, description="连接各阶段的入侵指标 (IOCs,如IP、Hash)。"
- )
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CYBER_KILL_CHAIN_CONSOLIDATED_PROMPT = (
- "你是一位数字取证专家和事件响应人员。请从事件报告中重建攻击时间线和杀伤链。\n\n"
- "提取规则:\n"
- "- 识别不同的攻击阶段作为节点。\n"
- "- 映射攻击在时间上和逻辑上的演进过程(边)。\n"
- "- 捕获每个阶段的具体战术、工具 (TTPs) 和时间戳。\n"
- "- 重点关注攻击者是如何移动以及在何时移动的。"
-)
-
-CYBER_KILL_CHAIN_NODE_PROMPT = (
- "识别所有可疑活动和网络入侵的不同阶段。\n\n"
- "提取规则:\n"
- "- 提取以行动为导向的阶段(例如:'投放载荷'、'建立C2')。\n"
- "- 识别与每个发现相关的工具和技术。\n"
- "- 在此阶段不要建立各阶段之间的连接。"
-)
-
-CYBER_KILL_CHAIN_EDGE_PROMPT = (
- "建立已识别攻击阶段之间的顺序和演进关系。\n\n"
- "提取规则:\n"
- "- 定义哪个阶段导致或促成了下一个阶段。\n"
- "- 识别连接这些阶段的入侵指标(IP、哈希等)。\n"
- "- 仅在已识别的列表阶段之间创建连接。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class IntrusionKillChainGraph(AutoGraph[AttackStageNode, KillChainLink]):
- """
- 用于重建网络攻击生命周期的模板。
-
- 适用于事件响应 (IR)、取证报告和威胁狩猎。
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> kill_chain = IntrusionKillChainGraph(llm_client=llm, embedder=embedder)
- >>> kill_chain.feed_text("攻击者通过鱼叉式网络钓鱼邮件获得了初始访问。")
- >>> print(kill_chain.nodes) # 获取攻击阶段
- >>> print(kill_chain.edges) # 获取演进链接
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化入侵杀伤链图谱模板。
-
- Args:
- llm_client (BaseChatModel): 用于杀伤链提取的大模型客户端。
- embedder (Embeddings): 用于阶段去重的嵌入模型。
- extraction_mode (str, optional): 'one_stage' 为联合提取,
- 'two_stage' 为分步提取。默认为 "one_stage"。
- chunk_size (int, optional): 单个文本块的最大字符数。默认为 2048。
- chunk_overlap (int, optional): 文本块之间的重叠字符数。默认为 256。
- max_workers (int, optional): 并行处理的工作线程数。默认为 10。
- verbose (bool, optional): 是否启用详细日记记录。默认为 False。
- **kwargs (Any): 传递给 AutoGraph 基类的其他参数。
- """
- super().__init__(
- node_schema=AttackStageNode,
- edge_schema=KillChainLink,
- node_key_extractor=lambda x: x.stage_name.strip().lower(),
- edge_key_extractor=lambda x: (
- f"{x.source.strip().lower()}->{x.target.strip().lower()}"
- ),
- nodes_in_edge_extractor=lambda x: (
- x.source.strip().lower(),
- x.target.strip().lower(),
- ),
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CYBER_KILL_CHAIN_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CYBER_KILL_CHAIN_NODE_PROMPT,
- prompt_for_edge_extraction=CYBER_KILL_CHAIN_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: AttackStageNode) -> str:
- return f"{node.stage_name}"
-
- def edge_label_extractor(edge: KillChainLink) -> str:
- return f"{edge.causality}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/cybersecurity/threat_intelligence_nexus.py b/hyperextract/templates/zh/cybersecurity/threat_intelligence_nexus.py
deleted file mode 100644
index f0f7c38..0000000
--- a/hyperextract/templates/zh/cybersecurity/threat_intelligence_nexus.py
+++ /dev/null
@@ -1,176 +0,0 @@
-from typing import List, Optional, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoHypergraph
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class ThreatActorNode(BaseModel):
- """
- 负责网络威胁的实体(个人、团体或国家级组织)。
- """
-
- actor_name: str = Field(
- description="别名或团体名称(例如:'APT28'、'Fancy Bear'、'Lazarus Group')。"
- )
- origin_country: Optional[str] = Field(None, description="疑似地理来源。")
- motivation: Optional[str] = Field(
- None,
- description="主要驱动因素:'间谍活动'、'经济利益'、'破坏'、'黑客行动主义'。",
- )
- typical_targets: List[str] = Field(
- default_factory=list, description="通常的目标行业或地区。"
- )
-
-
-class IntelligenceNexus(BaseModel):
- """
- 连接攻击者、特定基础设施和战役主题的超边。
- """
-
- participants: List[str] = Field(
- description="涉及的实体:攻击者名称、工具名称、基础设施(IP/域名)、受害组织。"
- )
- campaign_name: str = Field(
- description="已知的行动名称(例如:'SolarWinds 供应链攻击')。"
- )
- shared_ttp: str = Field(
- description="连接这些不同实体的共同战术、技术或过程(TTP)。"
- )
- confidence_level: str = Field("中", description="分析置信度:'高'、'中'、'低'。")
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CYBER_NEXUS_CONSOLIDATED_PROMPT = (
- "你是一位战略威胁情报分析师。请映射攻击者、基础设施及其战役之间的高层级关联。\n\n"
- "提取规则:\n"
- "- 识别威胁攻击者及其别名作为节点。\n"
- "- 创建超边 (Nexuses),在命名的战役下,将攻击者与特定工具和基础设施关联起来。\n"
- "- 重点关注揭示协同行动的多点连接,而非简单的二元链接。\n"
- "- 突出显示统一该 Nexus 的共享 TTPs。"
-)
-
-CYBER_NEXUS_NODE_PROMPT = (
- "识别威胁攻击者、恶意软件家族和恶意基础设施实体。\n\n"
- "提取规则:\n"
- "- 提取攻击者别名和团体名称。\n"
- "- 识别攻击动机和目标行业。\n"
- "- 在此阶段不要建立战役关联。"
-)
-
-CYBER_NEXUS_EDGE_PROMPT = (
- "识别统一了攻击者、工具和目标的行动战役 (Campaigns) 或关联 (Nexuses)。\n\n"
- "提取规则:\n"
- "- 每个超边必须描述一个单一的战役或行动。\n"
- "- 将攻击者关联到在该行动中使用的特定基础设施和工具。\n"
- "- 仅在已识别的列表实体之间创建连接。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class ThreatIntelligenceNexusHypergraph(
- AutoHypergraph[ThreatActorNode, IntelligenceNexus]
-):
- """
- 用于战略威胁情报映射的模板。
-
- 适用于在复杂战役中连接基础设施、攻击者和恶意软件。
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> nexus = ThreatIntelligenceNexusHypergraph(llm_client=llm, embedder=embedder)
- >>> nexus.feed_text("APT28 在 'Fancy-Phish' 行动中使用了 X-Tunnel。")
- >>> print(nexus.nodes) # 获取威胁攻击者
- >>> print(nexus.edges) # 获取战役超边
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- extraction_mode: str = "one_stage",
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化威胁情报关联超图模板。
-
- Args:
- llm_client (BaseChatModel): 用于关联提取的大模型客户端。
- embedder (Embeddings): 用于攻击者去重的嵌入模型。
- extraction_mode (str, optional): 'one_stage' 为联合提取,
- 'two_stage' 为分步提取。默认为 "one_stage"。
- chunk_size (int, optional): 单个文本块的最大字符数。默认为 2048。
- chunk_overlap (int, optional): 文本块之间的重叠字符数。默认为 256。
- max_workers (int, optional): 并行处理的工作线程数。默认为 10。
- verbose (bool, optional): 是否启用详细日记记录。默认为 False。
- **kwargs (Any): 传递给 AutoHypergraph 基类的其他参数。
- """
- super().__init__(
- node_schema=ThreatActorNode,
- edge_schema=IntelligenceNexus,
- node_key_extractor=lambda x: x.actor_name.strip().lower(),
- edge_key_extractor=lambda x: (
- f"{x.campaign_name.strip().lower()}_linking_{sorted(x.participants)}"
- ),
- nodes_in_edge_extractor=lambda x: [
- p.strip().lower() for p in x.participants
- ],
- llm_client=llm_client,
- embedder=embedder,
- extraction_mode=extraction_mode,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CYBER_NEXUS_CONSOLIDATED_PROMPT,
- prompt_for_node_extraction=CYBER_NEXUS_NODE_PROMPT,
- prompt_for_edge_extraction=CYBER_NEXUS_EDGE_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_nodes_for_search: int = 3,
- top_k_edges_for_search: int = 3,
- top_k_nodes_for_chat: int = 3,
- top_k_edges_for_chat: int = 3,
- ) -> None:
- """
- Visualize the graph using OntoSight.
-
- Args:
- top_k_nodes_for_search (int): Number of nodes to retrieve for search context. Default 3.
- top_k_edges_for_search (int): Number of edges to retrieve for search context. Default 3.
- top_k_nodes_for_chat (int): Number of nodes to retrieve for chat context. Default 3.
- top_k_edges_for_chat (int): Number of edges to retrieve for chat context. Default 3.
- """
- def node_label_extractor(node: ThreatActorNode) -> str:
- return f"{node.actor_name}"
-
- def edge_label_extractor(edge: IntelligenceNexus) -> str:
- return f"{edge.campaign_name}"
-
- super().show(
- node_label_extractor=node_label_extractor,
- edge_label_extractor=edge_label_extractor,
- top_k_nodes_for_search=top_k_nodes_for_search,
- top_k_edges_for_search=top_k_edges_for_search,
- top_k_nodes_for_chat=top_k_nodes_for_chat,
- top_k_edges_for_chat=top_k_edges_for_chat,
- )
diff --git a/hyperextract/templates/zh/cybersecurity/vulnerability_catalog.py b/hyperextract/templates/zh/cybersecurity/vulnerability_catalog.py
deleted file mode 100644
index 1c93e69..0000000
--- a/hyperextract/templates/zh/cybersecurity/vulnerability_catalog.py
+++ /dev/null
@@ -1,120 +0,0 @@
-from typing import Optional, List, Any
-from pydantic import BaseModel, Field
-from langchain_core.language_models.chat_models import BaseChatModel
-from langchain_core.embeddings import Embeddings
-from hyperextract.types import AutoSet
-
-# ==============================================================================
-# 1. Schema Definitions
-# ==============================================================================
-
-
-class VulnerabilityEntry(BaseModel):
- """
- 系统中识别出的特定安全漏洞或弱点。
- """
-
- vulnerability_id: str = Field(
- description="唯一标识符(例如:CVE-2023-1234, CWE-79)。"
- )
- title: str = Field(description="漏洞的简短描述性名称。")
- severity: str = Field(
- description="严重程度:'危急'、'高危'、'中危'、'低危'、'提示'。"
- )
- cvss_score: Optional[float] = Field(
- None, description="CVSS 基础评分(0.0 到 10.0)。"
- )
- affected_components: List[str] = Field(
- default_factory=list, description="受影响的软件或硬件组件。"
- )
- remediation: Optional[str] = Field(None, description="建议的修复或缓解措施。")
-
-
-# ==============================================================================
-# 2. Prompts
-# ==============================================================================
-
-CYBER_VULN_PROMPT = (
- "你是一位资深安全分析师和漏洞研究员。请从威胁报告、安全公告或扫描结果中提取并整合安全漏洞。\n\n"
- "提取规则:\n"
- "- 通过 CVE/CWE ID 或名称识别唯一的漏洞。\n"
- "- 准确捕获严重级别和 CVSS 评分。\n"
- "- 将同一漏洞的多次提及合并为单个条目,并汇总受影响组件和修复建议的信息。"
-)
-
-# ==============================================================================
-# 3. Template Class
-# ==============================================================================
-
-
-class VulnerabilityUnifiedCatalog(AutoSet[VulnerabilityEntry]):
- """
- 用于构建去重后的安全漏洞目录的模板。
-
- 适用于处理多个安全公告并创建用于补丁管理的清单。
-
- Example:
- >>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
- >>> catalog = VulnerabilityUnifiedCatalog(llm_client=llm, embedder=embedder)
- >>> catalog.feed_text("CVE-2024-0001 是内核中的一个危急 RCE 漏洞。")
- >>> catalog.feed_text("该内核 RCE (CVE-2024-0001) 需要立即修复。")
- >>> print(catalog.items) # CVE-2024-0001 的重复条目将被合并
- """
-
- def __init__(
- self,
- llm_client: BaseChatModel,
- embedder: Embeddings,
- *,
- chunk_size: int = 2048,
- chunk_overlap: int = 256,
- max_workers: int = 10,
- verbose: bool = False,
- **kwargs: Any,
- ):
- """
- 初始化安全漏洞目录模板。
-
- Args:
- llm_client (BaseChatModel): 用于漏洞提取的大模型客户端。
- embedder (Embeddings): 用于条目去重的嵌入模型。
- chunk_size (int, optional): 单个文本块的最大字符数。默认为 2048。
- chunk_overlap (int, optional): 文本块之间的重叠字符数。默认为 256。
- max_workers (int, optional): 并行处理的工作线程数。默认为 10。
- verbose (bool, optional): 是否启用详细日记记录。默认为 False。
- **kwargs (Any): 传递给 AutoSet 基类的其他参数。
- """
- super().__init__(
- item_schema=VulnerabilityEntry,
- key_extractor=lambda x: x.vulnerability_id.strip().upper(),
- llm_client=llm_client,
- embedder=embedder,
- chunk_size=chunk_size,
- chunk_overlap=chunk_overlap,
- max_workers=max_workers,
- verbose=verbose,
- prompt=CYBER_VULN_PROMPT,
- **kwargs,
- )
-
- def show(
- self,
- *,
- top_k_for_search: int = 3,
- top_k_for_chat: int = 3,
- ) -> None:
- """
- Visualize the collection using OntoSight.
-
- Args:
- top_k_for_search (int): Number of items to retrieve for search context. Default 3.
- top_k_for_chat (int): Number of items to retrieve for chat context. Default 3.
- """
- def item_label_extractor(item: VulnerabilityEntry) -> str:
- return f"{item.title}"
-
- super().show(
- item_label_extractor=item_label_extractor,
- top_k_for_search=top_k_for_search,
- top_k_for_chat=top_k_for_chat,
- )
diff --git a/hyperextract/templates/zh/fashion/__init__.py b/hyperextract/templates/zh/fashion/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/marketing/__init__.py b/hyperextract/templates/zh/marketing/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/music/__init__.py b/hyperextract/templates/zh/music/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/politics/__init__.py b/hyperextract/templates/zh/politics/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/real_estate/__init__.py b/hyperextract/templates/zh/real_estate/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/recruitment/__init__.py b/hyperextract/templates/zh/recruitment/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/software/__init__.py b/hyperextract/templates/zh/software/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/sports/__init__.py b/hyperextract/templates/zh/sports/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/supply_chain/__init__.py b/hyperextract/templates/zh/supply_chain/__init__.py
deleted file mode 100644
index e69de29..0000000
diff --git a/hyperextract/templates/zh/travel/__init__.py b/hyperextract/templates/zh/travel/__init__.py
deleted file mode 100644
index e69de29..0000000