Skip to content

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

feat: 优化概念图、知识图谱和逻辑图的模型定义,增强字段描述和格式化

yifanfeng97f2824d2f8914
diff --git a/hyperextract/templates/en/general/concept_map.py b/hyperextract/templates/en/general/concept_map.py
index 78f2dac..bbdeae4 100644
--- a/hyperextract/templates/en/general/concept_map.py
+++ b/hyperextract/templates/en/general/concept_map.py
@@ -8,21 +8,32 @@ from hyperextract.types import AutoGraph
# 1. Schema Definitions
# ==============================================================================
+
class Concept(BaseModel):
"""A conceptual node representing a term, idea, or theory."""
+
term: str = Field(description="The technical term or concept name.")
- definition: str = Field(description="A formal definition or explanation of the concept.")
- examples: Optional[List[str]] = Field(default_factory=list, description="Examples that illustrate this concept.")
- attributes: Optional[List[str]] = Field(default_factory=list, description="Key properties or attributes of the concept.")
+ definition: str = Field(
+ description="A formal definition or explanation of the concept."
+ )
+ examples: Optional[List[str]] = Field(
+ default_factory=list, description="Examples that illustrate this concept."
+ )
+ attributes: Optional[List[str]] = Field(
+ default_factory=list, description="Key properties or attributes of the concept."
+ )
+
class ConceptRelation(BaseModel):
"""A semantic relationship between two concepts."""
+
source: str = Field(description="The source concept.")
target: str = Field(description="The target concept.")
relation_type: str = Field(
description="Type of relationship (e.g., is-a, part-of, related-to, used-for, instance-of)."
)
+
# ==============================================================================
# 2. Prompts
# ==============================================================================
@@ -52,13 +63,14 @@ CONCEPT_MAP_EDGE_PROMPT = (
# 3. Template Class
# ==============================================================================
+
class ConceptMap(AutoGraph[Concept, ConceptRelation]):
"""
A template for building concept maps and taxonomies.
-
+
Ideal for structured learning, technical documentation, and glossary extraction.
It focuses on definitions, semantic categories, and hierarchical relationships.
-
+
Example:
>>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
>>> llm = ChatOpenAI(model="gpt-4o")
@@ -70,6 +82,7 @@ class ConceptMap(AutoGraph[Concept, ConceptRelation]):
>>> cm.feed_text(text)
>>> print(cm.edges) # Output shows: Machine Learning --(is a)--> Artificial Intelligence
"""
+
def __init__(
self,
llm_client: BaseChatModel,
@@ -80,7 +93,7 @@ class ConceptMap(AutoGraph[Concept, ConceptRelation]):
chunk_overlap: int = 256,
max_workers: int = 10,
verbose: bool = False,
- **kwargs: Any
+ **kwargs: Any,
):
"""
Initialize the ConceptMap template.
@@ -101,7 +114,9 @@ class ConceptMap(AutoGraph[Concept, ConceptRelation]):
node_schema=Concept,
edge_schema=ConceptRelation,
node_key_extractor=lambda x: x.term.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.relation_type.lower()})-->{x.target.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,
@@ -113,8 +128,9 @@ class ConceptMap(AutoGraph[Concept, ConceptRelation]):
prompt=CONCEPT_MAP_PROMPT,
prompt_for_node_extraction=CONCEPT_MAP_NODE_PROMPT,
prompt_for_edge_extraction=CONCEPT_MAP_EDGE_PROMPT,
- **kwargs
+ **kwargs,
)
+
def show(
self,
*,
@@ -125,19 +141,20 @@ class ConceptMap(AutoGraph[Concept, ConceptRelation]):
) -> 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: Concept) -> str:
- return f"{ node.term }"
-
+ return f"{node.term}"
+
def edge_label_extractor(edge: ConceptRelation) -> str:
- return f"{ edge.source }"
-
+ return f"{edge.relation_type}"
+
super().show(
node_label_extractor=node_label_extractor,
edge_label_extractor=edge_label_extractor,
diff --git a/hyperextract/templates/en/general/knowledge_graph.py b/hyperextract/templates/en/general/knowledge_graph.py
index fb59066..0cbb637 100644
--- a/hyperextract/templates/en/general/knowledge_graph.py
+++ b/hyperextract/templates/en/general/knowledge_graph.py
@@ -8,8 +8,10 @@ from hyperextract.types import AutoGraph
# 1. Schema Definitions
# ==============================================================================
+
class Entity(BaseModel):
"""A general entity representing a person, organization, location, or object."""
+
name: str = Field(description="The primary name of the entity.")
category: str = Field(
description="Category of the entity (e.g., Person, Organization, Location, Event, Object)."
@@ -18,12 +20,19 @@ class Entity(BaseModel):
description="A concise summary of the entity and its role in the text."
)
+
class Relation(BaseModel):
"""A factual relationship between two entities."""
+
source: str = Field(description="The name of the source entity.")
target: str = Field(description="The name of the target entity.")
- relation: str = Field(description="The type of relationship (e.g., Works for, Located in, Born in, Acquired).")
- details: Optional[str] = Field(description="Additional context or details about this relationship.")
+ relation: str = Field(
+ description="The type of relationship (e.g., Works for, Located in, Born in, Acquired)."
+ )
+ details: Optional[str] = Field(
+ description="Additional context or details about this relationship."
+ )
+
# ==============================================================================
# 2. Prompts
@@ -54,13 +63,14 @@ KNOWLEDGE_GRAPH_EDGE_PROMPT = (
# 3. Template Class
# ==============================================================================
+
class KnowledgeGraph(AutoGraph[Entity, Relation]):
"""
A foundational knowledge graph template for factual extraction.
-
- This template is optimized for extracting entities (People, Places, Organizations)
+
+ This template is optimized for extracting entities (People, Places, Organizations)
and their factual interactions from news articles, biographies, and encyclopedic texts.
-
+
Example:
>>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
>>> llm = ChatOpenAI(model="gpt-4o")
@@ -72,6 +82,7 @@ class KnowledgeGraph(AutoGraph[Entity, Relation]):
>>> kg.feed_text(text)
>>> print(kg.nodes) # Output: [Entity(name='Steve Jobs', ...), Entity(name='Apple', ...)]
"""
+
def __init__(
self,
llm_client: BaseChatModel,
@@ -82,7 +93,7 @@ class KnowledgeGraph(AutoGraph[Entity, Relation]):
chunk_overlap: int = 256,
max_workers: int = 10,
verbose: bool = False,
- **kwargs: Any
+ **kwargs: Any,
):
"""
Initialize the KnowledgeGraph template.
@@ -103,7 +114,9 @@ class KnowledgeGraph(AutoGraph[Entity, Relation]):
node_schema=Entity,
edge_schema=Relation,
node_key_extractor=lambda x: x.name.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--[{x.relation.lower()}]-->{x.target.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,
@@ -115,8 +128,9 @@ class KnowledgeGraph(AutoGraph[Entity, Relation]):
prompt=KNOWLEDGE_GRAPH_PROMPT,
prompt_for_node_extraction=KNOWLEDGE_GRAPH_NODE_PROMPT,
prompt_for_edge_extraction=KNOWLEDGE_GRAPH_EDGE_PROMPT,
- **kwargs
+ **kwargs,
)
+
def show(
self,
*,
@@ -127,20 +141,21 @@ class KnowledgeGraph(AutoGraph[Entity, Relation]):
) -> 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: Entity) -> str:
- info = f" ({ node.category })" if getattr(node, "category", None) else ""
- return f"{ node.name }{info}"
-
+ info = f" ({node.category})" if getattr(node, "category", None) else ""
+ return f"{node.name}{info}"
+
def edge_label_extractor(edge: Relation) -> str:
- return f"{ edge.source }"
-
+ return f"{edge.relation}"
+
super().show(
node_label_extractor=node_label_extractor,
edge_label_extractor=edge_label_extractor,
diff --git a/hyperextract/templates/en/general/logic_graph.py b/hyperextract/templates/en/general/logic_graph.py
index d864794..737c3cf 100644
--- a/hyperextract/templates/en/general/logic_graph.py
+++ b/hyperextract/templates/en/general/logic_graph.py
@@ -1,4 +1,4 @@
-from typing import List, Optional, Any
+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
@@ -8,22 +8,29 @@ from hyperextract.types import AutoGraph
# 1. Schema Definitions
# ==============================================================================
+
class LogicNode(BaseModel):
"""A node representing a claim, evidence, or premise in an argument."""
+
statement: str = Field(description="The central claim, fact, or observation.")
node_type: str = Field(
description="Type of node: Claim (main point), Evidence (supporting data), Premise (underlying assumption)."
)
- source_attribution: Optional[str] = Field(description="Where or who this logic point originates from.")
+ source_attribution: Optional[str] = Field(
+ description="Where or who this logic point originates from."
+ )
+
class LogicRelation(BaseModel):
"""The logical connection between two statements."""
+
source: str = Field(description="The source statement.")
target: str = Field(description="The target statement.")
inference: str = Field(
description="The logical link: Supports, Contradicts, Proves, Leads to, Explains."
)
+
# ==============================================================================
# 2. Prompts
# ==============================================================================
@@ -50,14 +57,15 @@ LOGIC_GRAPH_EDGE_PROMPT = (
# 3. Template Class
# ==============================================================================
+
class LogicGraph(AutoGraph[LogicNode, LogicRelation]):
"""
A template for analyzing reasoning, arguments, and causal chains.
-
- This template is designed to map out the logical structure of analytical reports,
- scientific papers, or debate transcripts. It captures claims, evidence, and
+
+ This template is designed to map out the logical structure of analytical reports,
+ scientific papers, or debate transcripts. It captures claims, evidence, and
the logical links (Support/Contradict) between them.
-
+
Example:
>>> from langchain_openai import ChatOpenAI, OpenAIEmbeddings
>>> llm = ChatOpenAI(model="gpt-4o")
@@ -69,6 +77,7 @@ class LogicGraph(AutoGraph[LogicNode, LogicRelation]):
>>> lg.feed_text(text)
>>> print(lg.edges) # Output shows: Carbon levels rising --(Supports)--> Climate is warming
"""
+
def __init__(
self,
llm_client: BaseChatModel,
@@ -79,7 +88,7 @@ class LogicGraph(AutoGraph[LogicNode, LogicRelation]):
chunk_overlap: int = 256,
max_workers: int = 10,
verbose: bool = False,
- **kwargs: Any
+ **kwargs: Any,
):
"""
Initialize the LogicGraph template.
@@ -100,7 +109,9 @@ class LogicGraph(AutoGraph[LogicNode, LogicRelation]):
node_schema=LogicNode,
edge_schema=LogicRelation,
node_key_extractor=lambda x: x.statement.strip(),
- edge_key_extractor=lambda x: f"{x.source.strip()}--({x.inference.lower()})-->{x.target.strip()}",
+ edge_key_extractor=lambda x: (
+ f"{x.source.strip()}--({x.inference.lower()})-->{x.target.strip()}"
+ ),
nodes_in_edge_extractor=lambda x: (x.source.strip(), x.target.strip()),
llm_client=llm_client,
embedder=embedder,
@@ -112,8 +123,9 @@ class LogicGraph(AutoGraph[LogicNode, LogicRelation]):
prompt=LOGIC_GRAPH_PROMPT,
prompt_for_node_extraction=LOGIC_GRAPH_NODE_PROMPT,
prompt_for_edge_extraction=LOGIC_GRAPH_EDGE_PROMPT,
- **kwargs
+ **kwargs,
)
+
def show(
self,
*,
@@ -124,19 +136,20 @@ class LogicGraph(AutoGraph[LogicNode, LogicRelation]):
) -> 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: LogicNode) -> str:
- return f"{ node.statement }"
-
+ return f"{node.statement}"
+
def edge_label_extractor(edge: LogicRelation) -> str:
- return f"{ edge.source }"
-
+ return f"{edge.inference}"
+
super().show(
node_label_extractor=node_label_extractor,
edge_label_extractor=edge_label_extractor,