Top-Level

The Basic Formal Ontology (BFO) provides the foundational framework for the Naas Ontology. BFO is a top-level ontology that systematically categorizes all entities that exist or can exist.

Ontology Engineering Core Principle

Ontology engineering aims to systematically disambiguate data to simultaneously advance interoperability and information quality. This distinguishes it from data modeling, knowledge representation, and taxonomy creation through its focus on formal precision and systematic disambiguation.

BFO High-Level Structure

BFO divides all entities into two fundamental categories:

Continuants - entities that persist through time
Occurrents - entities that happen in time

Systematic Ontology Engineering

Professional ontology engineering requires a disciplined approach to avoid the chaos of ad-hoc modeling. The 7 buckets strategy provides this discipline by ensuring systematic coverage of any domain through seven fundamental questions. This method prevents gaps, reduces overlap, and maintains consistency across different modeling efforts.

The 7 Buckets Strategy

BFO provides systematic rules for modeling any domain through seven fundamental questions:

Bucket	Question	Class	Relation
1	What materials exist?	Material Entity	-
2	What qualities do they have?	Quality	inheres in
3	What could they do?	Realizable Entity	inheres in, has material basis
4	What do they actually do?	Process	participates in
5	Where are they located?	Immaterial Entity	located in
6	When do they exist?	Temporal Region	exists at, datatype property
7	What information describes 1-6?	Generically Dependent Continuant	generically depends on

Additional relation: Realizables to processes → has realization

Rules of Thumb

When identifying relations, describe:

Qualities to material entities, i.e. inheres in
Realizables to material entities, i.e. inheres in, has material basis
Processes to material entities, i.e participates in
Realizables to processes, i.e. has realization
Immaterial location of material entity, i.e. located in
When any such entities exist, i.e. exists at, datatype property
When any such entities carry information, e.g. generically depends on

When identifying classes, describe:

Material entities within scope, i.e. Material Entity
Qualities these material entities have, i.e. Quality
What these material entities could do, i.e. Realizable Entity
What these material entities actually do, i.e. Process
Where these entities and boundaries are located, i.e. Immaterial Entity
When these entities exist, i.e. Temporal Region
Information about 1-6, i.e. Generically Dependent Continuant

Implementation Example: Naas Ontology

The following examples demonstrate how the 7 buckets strategy is applied to model AI systems in the Naas domain. Each code block shows the systematic classification of domain entities according to BFO principles, with proper class hierarchies and relation patterns.

1. Material Entities (WHAT/WHO)

The code defines AISystem and AIModelInstance as subclasses of bfo:BFO_0000040 (Material Entity). Notice how each class includes rdfs:label, skos:definition, and skos:example to provide human-readable descriptions and concrete instances:

@prefix bfo: <http://purl.obolibrary.org/obo/> .
@prefix abi: <http://ontology.naas.ai/abi/> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix skos: <http://www.w3.org/2004/02/skos/core#> .

# AI Systems as material entities
abi:AISystem a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000040 ; # Material Entity
    rdfs:label "AI System"@en ;
    skos:definition "A material entity that implements artificial intelligence capabilities."@en ;
    skos:example "ChatGPT server infrastructure, Claude API endpoints, Gemini model instances."@en .

abi:AIModelInstance a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000040 ; # Material Entity
    rdfs:label "AI Model Instance"@en ;
    skos:definition "A material entity that is a specific instantiation of an AI model running on hardware."@en .

2. Qualities (HOW-IT-IS)

The code defines three quality classes: IntelligenceScore, ResponseSpeed, and CostPerToken, all as subclasses of bfo:BFO_0000019 (Quality). Each represents a measurable property that can inhere in AI systems:

# Performance qualities of AI systems
abi:IntelligenceScore a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000019 ; # Quality
    rdfs:label "Intelligence Score"@en ;
    skos:definition "A quality that measures the cognitive capability of an AI system."@en .

abi:ResponseSpeed a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000019 ; # Quality
    rdfs:label "Response Speed"@en ;
    skos:definition "A quality that measures how quickly an AI system responds to queries."@en .

abi:CostPerToken a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000019 ; # Quality
    rdfs:label "Cost Per Token"@en ;
    skos:definition "A quality that measures the financial cost per token processed."@en .

3. Realizable Entities (WHY-POTENTIAL)

The code defines AgentRole and CognitiveCapability as subclasses of bfo:BFO_0000017 (Realizable Entity). Note the skos:example showing specific instances like "Primary responder" and "Truth-seeking analysis":

# Agent roles and capabilities
abi:AgentRole a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000017 ; # Realizable Entity
    rdfs:label "Agent Role"@en ;
    skos:definition "A realizable entity that defines the specialized function or role of an AI agent."@en ;
    skos:example "Primary responder, fallback handler, specialized task executor."@en .

abi:CognitiveCapability a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000017 ; # Realizable Entity
    rdfs:label "Cognitive Capability"@en ;
    skos:definition "A realizable entity that represents a cognitive ability of an AI system."@en ;
    skos:example "Truth-seeking analysis, ethical reasoning, multimodal processing."@en .

4. Processes (HOW-IT-HAPPENS)

The code defines three process classes: AIProcessing, AgentRouting, and KnowledgeGraphConstruction, all as subclasses of bfo:BFO_0000015 (Process). Each represents activities that unfold over time:

# AI processing activities
abi:AIProcessing a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000015 ; # Process
    rdfs:label "AI Processing"@en ;
    skos:definition "A process in which an AI system analyzes input and generates output."@en .

abi:AgentRouting a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000015 ; # Process
    rdfs:label "Agent Routing"@en ;
    skos:definition "A process of selecting the appropriate AI agent based on cognitive requirements."@en .

abi:KnowledgeGraphConstruction a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000015 ; # Process
    rdfs:label "Knowledge Graph Construction"@en ;
    skos:definition "A process of building structured knowledge representations from data."@en .

5. Information Entities (HOW-WE-KNOW)

The code defines OntologyDefinition and SPARQLQuery as subclasses of bfo:BFO_0000031 (Generically Dependent Continuant). Both represent information structures that depend on some bearer for their existence:

# Information structures
abi:OntologyDefinition a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000031 ; # Generically Dependent Continuant
    rdfs:label "Ontology Definition"@en ;
    skos:definition "An information entity that specifies the structure and meaning of an ontology."@en .

abi:SPARQLQuery a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000031 ; # Generically Dependent Continuant
    rdfs:label "SPARQL Query"@en ;
    skos:definition "An information entity that represents a query in SPARQL language."@en .

6. Immaterial Entities (WHERE)

The code defines DataCenterLocation and CloudRegion as subclasses of bfo:BFO_0000141 (Immaterial Entity). These represent spatial boundaries and locations where material entities exist:

# Spatial locations and boundaries
abi:DataCenterLocation a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000141 ; # Immaterial Entity
    rdfs:label "Data Center Location"@en ;
    skos:definition "An immaterial entity that represents the spatial location of a data center."@en .

abi:CloudRegion a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000141 ; # Immaterial Entity
    rdfs:label "Cloud Region"@en ;
    skos:definition "An immaterial entity that represents a geographical region for cloud services."@en .

7. Temporal Regions (WHEN)

The code defines InferenceSession and ModelTrainingPeriod as subclasses of bfo:BFO_0000008 (Temporal Region). These represent time intervals during which processes occur:

# Temporal intervals
abi:InferenceSession a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000008 ; # Temporal Region
    rdfs:label "Inference Session"@en ;
    skos:definition "A temporal region during which an AI system performs inference."@en .

abi:ModelTrainingPeriod a owl:Class ;
    rdfs:subClassOf bfo:BFO_0000008 ; # Temporal Region
    rdfs:label "Model Training Period"@en ;
    skos:definition "A temporal region during which an AI model undergoes training."@en .

Next Steps

The BFO foundation provides the systematic basis for:

Mid-Level - Common Core Ontologies that build on BFO
Domain-Level - Specific domains grounded in BFO categories
Process-Centric Routing - AI routing based on BFO process analysis

BFO provides the rigorous ontological foundation that ensures consistency and interoperability across all levels of the Naas Ontology.

Ontology Engineering Core Principle​

BFO High-Level Structure​

Systematic Ontology Engineering​

The 7 Buckets Strategy​

Rules of Thumb​

When identifying relations, describe:​

When identifying classes, describe:​

Implementation Example: Naas Ontology​

1. Material Entities (WHAT/WHO)​

2. Qualities (HOW-IT-IS)​

3. Realizable Entities (WHY-POTENTIAL)​

4. Processes (HOW-IT-HAPPENS)​

5. Information Entities (HOW-WE-KNOW)​

6. Immaterial Entities (WHERE)​

7. Temporal Regions (WHEN)​

Next Steps​