W3C Community Group

Agent Ontology Alignment:
IDL Agent Ontology and HMAS Ontology Stack

Document type:
W3C Community Group Draft
Published:
8 March 2026
This version:
https://www.w3.org/community/aikr/
Editor:
Paola Di Maio, W3C AI Knowledge Representation CG
Community Group:
W3C AI Knowledge Representation Community Group (AI KR CG)

Copyright © 2026 the Contributors to the Agent Ontology Alignment specification, published by the AI Knowledge Representation Community Group under the W3C Community Contributor License Agreement (CLA). A human-readable summary is available.

Status: This report was published by the AI Knowledge Representation Community Group. It is not a W3C Standard nor is it on the W3C Standards Track. Please note that under the W3C Community Contributor License Agreement (CLA) there is a limited opt-out and other conditions apply. Learn more about W3C Community and Business Groups.

Abstract

This technical note describes a formal method for ontology alignment and provides a formal logical alignment between the IDL Agent Ontology (from the s-agent-comm repository) and the HMAS ontology stack -- comprising the Core, Interaction, and Regulation modules. The goal is to define semantic correspondences (subClassOf, equivalentClass, and property mappings) so that systems using both vocabularies can interoperate. This document presents the alignment method, a use-case application, and open questions for community review.

Status of This Document

This is a W3C Community Group Draft published by the W3C AI Knowledge Representation Community Group. It is submitted for review by CG members and invited experts. Comments are welcome via the group's public mailing list or GitHub repository.

The Open Questions section actively solicits feedback on method validity, mapping completeness, and generalisation to other ontologies.

Open Questions for Community Review

This draft is submitted to the W3C AI KR CG for review. Feedback is actively solicited on:

Q1. Is the four-step alignment method described in Section 1 valid? Can it be improved or generalised?
Q2. Is the resulting IDL/HMAS mapping satisfactory? Are there missing correspondences or incorrect subsumptions?
Q3. Can this method be applied to establish alignment between other ontologies in the W3C ecosystem?

Method

The alignment uses a semantic ontology matching approach combining conceptual analysis and logical mapping in OWL, summarised in four steps.

Ontology Scope Analysis

The conceptual scope of each ontology was first characterised:

This step identifies complementary modelling layers rather than competing ones.

Semantic Concept Matching

Three mapping types were used:

Equivalent classes
Used when two ontologies describe the same concept.
idl:Agent ≡ hmas:Agent idl:Action ≡ hmas:Action
Subclass relations
Used when one ontology defines a more specific specialisation.
idl:Intent ⊑ hmasint:InteractionGoal idl:Capability ⊑ hmas:Ability
Contextual specialisation
Used when a concept belongs to a broader system-level concept.
idl:ExecutionEnvironment ⊑ hmas:Environment idl:Delegation ⊑ hmasint:Interaction

Property Alignment

Equivalent properties
idl:hasCapability ≡ hmas:hasAbility idl:executesAction ≡ hmas:performs
Subproperty mappings
idl:delegatesTo ⊑ hmasint:initiatesInteraction

This preserves logical consistency while allowing reasoning engines to infer links.

Layered Ontology Integration

LayerOntology
RegulationHMAS Regulation [HMAS-REGULATION]
InteractionHMAS Interaction [HMAS-INTERACTION]
Core system entitiesHMAS Core [HMAS-CORE]
Agent internal semanticsIDL Ontology [IDL-AGENT-ONTOLOGY]

IDL describes internal agent semantics; HMAS describes ecosystem interaction and governance. This separation improves modularity and reuse.

Alignment Formalization

The mapping is expressed as an alignment function:

Φ : OIDL → OHMAS

where each mapping is represented using OWL axioms: owl:equivalentClass, rdfs:subClassOf, owl:equivalentProperty, rdfs:subPropertyOf.

Analysis Result

Conceptual Roles of the Two Ontology Families

IDL Agent Ontology

The IDL / s-agent-comm ontology [IDL-AGENT-ONTOLOGY] provides a vocabulary for agent identity, capabilities, intent, delegation, accountability, and execution context. Main modules: core (Agent, Action, Intent), capability, delegation, accountability, execution-context.

HMAS Ontologies

The HMAS stack models human-machine-agent ecosystems [HMAS-CORE] [HMAS-INTERACTION] [HMAS-REGULATION]: core (agents, artifacts, environments), interaction (communicative acts and interaction processes), regulation (norms, policies, obligations).

Ontology Alignment Strategy

Φ : OIDL → OHMAS

Mappings: ≡ equivalentClass  |  ⊑ subclassOf  |  → property alignment.

Core Class Alignments

Agent

IDL:Agent ≡ HMAS:Agent

idl:Agent owl:equivalentClass hmas:Agent .

Both represent autonomous actors capable of actions and interactions.

Capability

IDL:Capability ⊑ HMAS:Ability

idl:Capability rdfs:subClassOf hmas:Ability .

Action / Activity

IDL:Action ≡ HMAS:Action

idl:Action owl:equivalentClass hmas:Action .

Interaction Mappings

Intent

IDL:Intent ⊑ hmas:InteractionGoal

idl:Intent rdfs:subClassOf hmasint:InteractionGoal .

Delegation

IDL:Delegation ⊑ hmasint:Interaction

idl:Delegation rdfs:subClassOf hmasint:Interaction .

Message / Communicative Act

IDL:IntentMessage ⊑ hmasint:CommunicativeAct

idl:IntentMessage rdfs:subClassOf hmasint:CommunicativeAct .

Regulation Mappings

Accountability

IDL:AccountabilityEvent ⊑ hmasreg:NormEnforcement

idl:AccountabilityEvent rdfs:subClassOf hmasreg:NormEnforcement .

Responsibility Attribution

IDL:ResponsibilityAttribution ⊑ hmasreg:Obligation

idl:ResponsibilityAttribution rdfs:subClassOf hmasreg:Obligation .

Policy / Governance

IDL:SecurityContext ⊑ hmasreg:Policy

idl:SecurityContext rdfs:subClassOf hmasreg:Policy .

Execution Context Alignment

IDL:ExecutionEnvironment ⊑ HMAS:Environment

idl:ExecutionEnvironment rdfs:subClassOf hmas:Environment .

Property Alignment

IDL propertyHMAS propertyMapping type
idl:hasCapabilityhmas:hasAbilityequivalent
idl:delegatesTohmasint:initiatesInteractionsubProperty
idl:executesActionhmas:performsequivalent
idl:hasIntenthmasint:hasGoalequivalent
idl:hasCapability owl:equivalentProperty hmas:hasAbility . idl:delegatesTo rdfs:subPropertyOf hmasint:initiatesInteraction .

Resulting Layered Integration Model

+---------------------+ | HMAS Regulation | | (Norms / Policies) | +----------^----------+ | IDL Accountability | +--------------------+----------------------------+ | HMAS Interaction | | (Communicative acts) | +-------------^-----------------------------------+ | IDL Intent / Delegation | +-------------+--------------+ | HMAS Core | | Agent / Action / Env | +-------------^--------------+ | IDL Core + Capability + Execution Context

IDL -- internal agent semantics. HMAS -- system-level interactions and governance. Together: a complete socio-technical agent ecosystem model.

Example Integrated RDF

ex:agent1 a hmas:Agent , idl:Agent ; idl:hasCapability ex:textAnalysis ; idl:hasIntent ex:summarizeDocument ; hmas:performs ex:analysisAction . ex:analysisAction a hmas:Action , idl:Action . ex:summarizeDocument a idl:Intent ; hmasint:hasGoal ex:summaryGoal .

Summary

Formal correspondences established:

  1. Agent and Action -- equivalence
  2. Intent -- subclass of InteractionGoal
  3. Delegation -- subclass of Interaction
  4. Accountability -- subclass of NormEnforcement
  5. Execution context -- subclass of Environment

Clean modular alignment: IDL -- agent cognition and governance; HMAS -- interaction and socio-technical regulation.

References

[IDL-AGENT-ONTOLOGY]
IDL Agent Ontology, s-agent-comm, 2025. https://github.com/s-agent-comm/agent-idl
[HMAS-CORE]
Hypermedia Multi-Agent Systems Core Ontology, HyperAgents Project, 2021. https://purl.org/hmas/core
[HMAS-INTERACTION]
Hypermedia Multi-Agent Systems Interaction Ontology, HyperAgents Project, 2023. https://purl.org/hmas/interaction
[HMAS-REGULATION]
Hypermedia Multi-Agent Systems Regulation Ontology, HyperAgents Project, 2024. https://purl.org/hmas/regulation
[AIAO]
Paola Di Maio, AI Agent Ontology (AIAO), Figshare, 2025. https://doi.org/10.6084/m9.figshare.30760739