Skip to content

Plan, apply and roll back changes to a production ontology, with a blast radius report and a proof an auditor can re-check.

7.2 KB
Raw Blame

BORO Ontology Comparison: Hand-crafted vs AI-generated

Structural Metrics

Metric Hand-crafted AI-generated Delta
Total classes 43 13 +30
Entity+State pairs 9 3 +6
Unpaired entities (no State) 9 5 +4
ClassOf classes 10 0 +10
BoundingState subclasses 6 2 +4
Object properties 14 5 +9
Datatype properties 4 3 +1
Named individuals 21 8 +13
Triples (in namespace) 344 117 +227
Avg comment length (chars) 122 42 +80

Classes in Hand-crafted but NOT in AI-generated

These classes represent intermediate abstractions or speculative hierarchies that the AI considered unnecessary:

  • BuildingCondition
  • BuildingElement
  • BuildingElementState
  • ClassOfBuilding
  • ClassOfBuildingElement
  • ClassOfBuildingElementState
  • ClassOfBuildingState
  • ClassOfDwelling
  • ClassOfDwellingState
  • ClassOfFloor
  • ClassOfFloorState
  • ClassOfRoom
  • ClassOfRoomState
  • Door
  • DoorState
  • ElementMaterial
  • FloorLevelDesignation
  • FloorState
  • InspectionEvent
  • InstalledState
  • OccupancyType
  • OccupiedState
  • RemovedState
  • Roof
  • RoofState
  • VacatedState
  • Wall
  • WallState
  • Window
  • WindowState

Properties in Hand-crafted but NOT in AI-generated

These properties serve the intermediate classes that the AI omitted:

  • constructedBy
  • elementInRoom
  • hasCondition
  • hasElement
  • hasFloorLevel
  • hasMaterial
  • hasOccupancyType
  • inspectedEntity
  • roomInDwelling
  • roomCapacity

Entity+State Pair Comparison

Entity Hand-crafted has State? AI has State?
Building yes yes
BuildingCondition entity only ---
BuildingElement yes ---
ConstructionEvent entity only entity only
DemolitionEvent entity only entity only
Door yes ---
Dwelling yes yes
ElementMaterial entity only ---
Floor yes entity only
FloorLevelDesignation entity only ---
InspectionEvent entity only ---
OccupancyEvent entity only entity only
OccupancyType entity only ---
RenovationEvent entity only entity only
Roof yes ---
Room yes yes
Wall yes ---
Window yes ---

ClassOf Hierarchy Comparison

ClassOf Class Hand-crafted AI-generated
ClassOfBuilding yes ---
ClassOfBuildingElement yes ---
ClassOfBuildingElementState yes ---
ClassOfBuildingState yes ---
ClassOfDwelling yes ---
ClassOfDwellingState yes ---
ClassOfFloor yes ---
ClassOfFloorState yes ---
ClassOfRoom yes ---
ClassOfRoomState yes ---

Summary

  • Class reduction: 70% fewer classes (43 -> 13)
  • Triple reduction: 66% fewer triples (344 -> 117)
  • Comment verbosity: hand-crafted averages 122 chars/comment vs AI's 42 chars/comment
  • ClassOf overhead: hand-crafted has 10 ClassOf classes; AI has 0
  • BoundingState overhead: hand-crafted has 6 BoundingState subclasses; AI has 2

Key Differences Explained

1. Intermediate Entity Classes

The hand-crafted ontology creates a BuildingElement superclass with concrete subclasses (Wall, Roof, Window, Door), each with their own *State class. The AI version omits these entirely -- individual building components are modelled as part-of relationships from Building, classified via ies:similarEntity when needed.

2. ClassOf Powertype Hierarchies

The hand-crafted ontology creates dedicated ClassOfX and ClassOfXState for every entity type (Building, Room, Floor, Dwelling, BuildingElement). The AI version uses the existing ies:ClassOfEntity directly for classification instances, avoiding 10 extra classes.

3. BoundingState Proliferation

The hand-crafted version creates 6 BoundingState subclasses (ConstructedState, DemolishedState, OccupiedState, VacatedState, InstalledState, RemovedState). The AI version keeps only the 2 essential ones (ConstructedState, DemolishedState) and handles occupancy via events.

4. Comment Verbosity

Hand-crafted comments explain BORO theory in each definition. AI-generated comments state what the class is concisely.

5. Speculative Classification Classes

The hand-crafted version includes OccupancyType, BuildingCondition, ElementMaterial, and FloorLevelDesignation as intermediate classification hierarchies. The AI version omits these because they can be added when actually needed, following YAGNI principles.

Why AI-native Ontology Engineering Matters

Traditional ontology engineering treats completeness as a virtue. Every conceivable classification axis is modelled upfront, every entity gets the full BORO treatment (Entity + State + ClassOf + ClassOfState), and comments explain the methodology rather than the domain.

The result is an ontology that is architecturally correct but operationally heavy. In this comparison:

  • The hand-crafted version has 43 classes to cover a domain that the AI covers with 13 classes -- a 70% reduction.
  • The hand-crafted version produces 344 triples vs the AI's 117 -- 66% less data to store, query, and maintain.
  • The AI's comments average 42 characters vs 122 -- they say what something is rather than why BORO requires it.

AI-native ontology engineering does not abandon rigour. Both ontologies use the same IES4 upper ontology, the same 4D perdurantist patterns, and the same Entity+State temporal modelling. The AI simply applies these patterns where they deliver value, rather than everywhere they could be applied.

The practical benefit: an AI-generated ontology is easier to understand, faster to query, cheaper to maintain, and can be extended incrementally when new requirements emerge -- rather than trying to anticipate all possible future needs upfront.