Ontology-based Semantic Interaction for Internet of Agents

Abstract

This document specifies a normative semantic layer for agent-to-agent interaction in the Internet of Agents (IoA). The semantic layer provides a common ontology model and a JSON-LD serialization profile (RECOMMENDED), while allowing other RDF serializations for expressing capabilities, intents, tasks, and context. The document defines required classes and properties, a negotiation procedure, and alignment rules for heterogeneous ontologies. This layer is designed to be used by interaction and discovery protocols but does not define transport, session, or security protocols. It enables deterministic semantic interoperability across domains.¶

1. Introduction

The Internet of Agents (IoA) envisions large-scale collaboration among heterogeneous agents across domains, vendors, and execution environments. Current agent ecosystems often fail to interoperate because capability descriptions, task intents, and context semantics are expressed using incompatible vocabularies. This semantic mismatch blocks cross-domain discovery, reliable task decomposition, and predictable coordination.¶

This document defines an ontology-based semantic layer for agent interaction. The layer provides:¶

- A minimal ontology model for capabilities, intents, tasks, and context.¶

- A JSON-LD serialization profile (RECOMMENDED) for exchange over existing interaction protocols, with allowance for other RDF serializations.¶

- A negotiation and alignment procedure for heterogeneous ontologies.¶

The semantic layer complements existing interaction and discovery protocols. It does not define transport, session, or security protocols, and it does not mandate a specific discovery system. It provides deterministic semantics that those protocols can carry.¶

2. Conventions and Terminology

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119] and [RFC8174].¶

The following terms are used in this document:¶

Internet of Agents (IoA): An ecosystem and architecture enabling interoperable collaboration among heterogeneous agents across domains and environments.¶

Semantic Layer: A logical layer that standardizes meaning representation for agent interaction, independent of transport.¶

Ontology: A formal vocabulary of concepts and relations that can be used to describe capabilities, intents, tasks, and context.¶

Semantic Profile: A minimal, structured set of fields used to express interaction semantics.¶

Capability: A well-defined function or service that an agent can provide, described semantically.¶

Intent: A goal-oriented statement describing the desired outcome of an interaction.¶

Task: A unit of work with a lifecycle and status, linked to intent and capability.¶

Context: Constraints, assumptions, and environmental conditions relevant to a task or interaction.¶

Alignment: A mapping procedure between ontologies to reconcile equivalent or related concepts.¶

3. Scope and Non-Goals

This document specifies:¶

- A normative semantic ontology model.¶

- A JSON-LD serialization profile (RECOMMENDED) for interaction and discovery semantics, with allowance for other RDF serializations.¶

- A negotiation procedure for semantic compatibility.¶

- Alignment artifacts and conflict resolution rules.¶

This document does not specify:¶

- Transport, session, or security protocols.¶

- A specific discovery or registration protocol.¶

- A full ontology library for a specific industry domain.¶

4. Semantic Layer Architecture

The semantic layer sits above interaction protocols and below application logic. It provides a structured meaning representation that can be exchanged, validated, and mapped across heterogeneous agents. The semantic layer is protocol-agnostic and can be carried by existing agent-to-agent interaction mechanisms.¶

The layer enables:¶

- Capability alignment across different vocabularies.¶

- Intent clarity for task decomposition and coordination.¶

- Context sharing for constraints and assumptions.¶

For cross-domain deployments, a shared semantic knowledge base (e.g., ontology and concept registries with version history) MAY be used as a common reference to improve semantic consistency.¶

The semantic layer is illustrated in Figure 1.¶

        +------------------------------+
        |      Application Logic       |
        +------------------------------+
        |   Semantic Layer (this doc)  |
        +------------------------------+
        |   Interaction Protocols      |
        +------------------------------+
        |   Transport / Security       |
        +------------------------------+

Figure 1: Semantic Layer Position¶

5. Ontology Model (Normative)

The ontology model defines the minimal classes and properties that MUST be supported for interoperable interaction. This model is RECOMMENDED to be expressed using RDF/OWL concepts [RDF11-CONCEPTS] [OWL2-OVERVIEW].¶

RDF provides a graph-based, globally identifiable data model that is well-suited to cross-domain integration, incremental extension, and ontology alignment. It enables explicit semantics, constraint validation, and reasoning using mature tooling, which improves interoperability, verifiability, and long-term reuse.¶

   @prefix ioa: <https://ex.org/ioa/semantic#> .

   @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .

   ioa:Domain   a rdfs:Class .
   ioa:Concept  a rdfs:Class .
   ioa:inDomain a rdfs:Property .
   ioa:abbr     a rdfs:Property .
   ioa:preferredAbbr a rdfs:Property .
   ioa:profileContext a rdfs:Property .

   @prefix ioa: <https://ex.org/ioa/semantic#> .
   @prefix cs:  <https://ex.org/onto/cs#> .

   cs:ComputerScience  a ioa:Domain .
   cs:LargeLanguageModel  a ioa:Concept ;
     ioa:inDomain cs:ComputerScience ;
     ioa:abbr "LLM" .

   cs:TechTranslation  a ioa:Capability .
   cs:TranslateTechDoc a ioa:Intent ;
     ioa:requiresCapability cs:TechTranslation ;
     ioa:appliesToDomain cs:ComputerScience .

   @prefix ioa: <https://ex.org/ioa/semantic#> .
   @prefix law: <https://ex.org/onto/law#> .

   law:Law  a ioa:Domain .
   law:MasterOfLaws  a ioa:Concept ;
     ioa:inDomain law:Law ;
     ioa:abbr "LLM" ;
     ioa:preferredAbbr "LL.M." .

6. JSON-LD Serialization Profile (Normative)

JSON-LD is the RECOMMENDED serialization for semantic payloads [JSON-LD]. Other semantically equivalent representations MAY be used if they preserve the required fields and deterministic interpretation. Other RDF serializations (Turtle, RDF/XML) MAY be supported for offline exchange.¶

   {
     "@context": [
       "https://ex.org/ctx",
       { "cs": "https://ex.org/onto/cs#" }
     ],
     "type": "ioa:SemanticProfile",
     "ontologyId": "https://ex.org/onto/agent-core",
     "ontologyVersion": "1.0",
     "capabilityRefs": ["cs:TechTranslation"],
     "intentRef": "cs:TranslateTechDoc",
     "profileContext": {
       "domain": "cs:ComputerScience",
       "termBindings": [
         {
           "term": "LLM",
           "sense": "cs:LargeLanguageModel"
         }
       ]
     }
   }

   {
     "@context": [
       "https://ex.org/ctx",
       { "cs": "https://ex.org/onto/cs#" }
     ],
     "type": "ioa:Capability",
     "id": "cs:TechTranslation",
     "label": "Technical Document Translation",
     "inputs": ["text/plain"],
     "outputs": ["text/plain"]
   }

7. Semantic Negotiation and Alignment

8. Discovery and Capability Semantics

This semantic layer can be used by discovery systems to match capability requirements against registered capabilities. Agents SHOULD advertise capabilities using ioa:Capability objects and include semantic profiles in discovery responses. The discovery protocol itself is out of scope; for example interactions see [IoA-Task].¶

9. Interaction Semantics

Interaction messages SHOULD carry a semantic envelope. Agents MUST ensure that semantic profiles associated with a task remain stable for the lifetime of the task unless renegotiated. If semantic renegotiation occurs, a new profile MUST be recorded with the task.¶

10. Mapping to ACPs/ADP/AIP

This semantic layer is intended to complement the Agent Collaboration Protocols (ACPs) architecture [ACPs-ARC] and can be mapped as follows:¶

- ADP (Agent Discovery Protocol): use SemanticProfile to represent capability requirements and results for semantic matching.¶

- AIP (Agent Interaction Protocol): attach SemanticProfile to interaction messages to make task intent and context explicit.¶

- Identity/authentication components: no changes required; semantic profiles are orthogonal to identity and authentication.¶

This mapping allows ACPs deployments to adopt semantic interoperability without modifying transport or security components.¶

This semantic layer also complements the DMSC infrastructure architecture draft [INF-ARCH] by providing semantic descriptions for capability directories (Section 6.1), discovery matching (Section 6.2), and semantic routing inputs (Section 6.3).¶

11. Example Flow and Examples

   {
     "@context": [
       "https://ex.org/ctx",
       { "clinical": "https://y.org/onto/clinical#" }
     ],
     "type": "ioa:SemanticProfile",
     "ontologyId": "https://y.org/onto/clinical",
     "ontologyVersion": "2.1",
     "capabilityRefs": ["clinical:note-summarize"],
     "intentRef": "clinical:summary-intent",
     "alignmentRef": "https://ex.org/align/X-Y-2026"
   }

   {
     "header": {
       "task_id": "t-1038",
       "message_type": "task_request"
     },
     "semantic": {
       "@context": [
         "https://ex.org/ctx",
         { "cs": "https://ex.org/onto/cs#" }
       ],
       "type": "ioa:SemanticProfile",
       "ontologyId": "https://ex.org/onto/agent-core",
       "ontologyVersion": "1.0",
       "capabilityRefs": ["cs:TechTranslation"],
       "intentRef": "cs:TranslateTechDoc",
        "profileContext": {
         "domain": "cs:ComputerScience",
         "sourceLang": "zh",
         "targetLang": "en",
         "termBindings": [
           {
             "term": "LLM",
             "sense": "cs:LargeLanguageModel"
           }
         ]
       }
     },
     "payload": {
       "input": "..."
     }
   }

   {
     "semantic": {
       "@context": [
         "https://ex.org/ctx",
         { "cs": "https://ex.org/onto/cs#" },
         { "law": "https://ex.org/onto/law#" }
       ],
       "type": "ioa:SemanticProfile",
       "ontologyId": "https://ex.org/onto/agent-core",
       "ontologyVersion": "1.0",
       "intentRef": "cs:TranslateTechDoc",
       "profileContext": {
         "domain": "cs:ComputerScience",
         "termBindings": [
           {
             "term": "LLM",
             "sense": "law:MasterOfLaws"
           }
         ]
       }
     },
     "validation": {
       "status": "reject",
       "reason": "termBinding sense conflicts with domain"
     }
   }

   {
     "sourceOntologyId": "https://x.org/onto/rad",
     "targetOntologyId": "https://y.org/onto/clinical",
     "mappingRules": [
       {
         "from": "rad:report-summary",
         "to": "clinical:note-summarize"
       }
     ],
     "confidence": 0.91
   }

12. Compliance Requirements

An implementation compliant with this specification MUST:¶

- Support the core classes and properties in Section 5.¶

- Produce and parse semantic envelopes as specified in Section 6; JSON-LD support is RECOMMENDED.¶

- Perform semantic negotiation prior to task execution (Section 7).¶

An implementation MAY support extended domain ontologies.¶

13. Security Considerations

Semantic profiles may reveal sensitive capability and context information. Implementations SHOULD consider:¶

- Minimizing semantic disclosure (least-information profiles).¶

- Integrity protection of semantic profiles and alignment artifacts.¶

- Access control on ontology registries and mapping services.¶

- Policy-based filtering of sensitive capabilities in discovery.¶

If a shared semantic knowledge base is used, implementers SHOULD consider provenance, version control, and trust policies to prevent poisoning or unauthorized changes.¶

Alignment artifacts can be a target for poisoning or downgrade attacks. Implementations SHOULD validate alignment sources and require provenance for mappings used in critical tasks.¶

14. IANA Considerations

This document makes no request for IANA action.¶

15. Normative References

[RFC2119]

Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, 1997, <https://www.rfc-editor.org/info/rfc2119>.

[RFC8174]

Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, 2017, <https://www.rfc-editor.org/info/rfc8174>.

16. Informative References

[ACPs-ARC]

Liu, J., Yu, K., Li, K., and K. Chen, "Agent Collaboration Protocols Architecture for Internet of Agents", Work in Progress, Internet-Draft, draft-liu-dmsc-acps-arc-03, 2026, <https://datatracker.ietf.org/doc/draft-liu-dmsc-acps-arc/03/>.

[INF-ARCH]

Li, X. and A. Wang, "Dynamic Multi-agent Secured Collaboration Infrastructure Architecture", Work in Progress, Internet-Draft, draft-li-dmsc-inf-architecture-03, 2026, <https://datatracker.ietf.org/doc/draft-li-dmsc-inf-architecture/03/>.

[IoA-Task]

Yang, C., "Internet of Agents Task Protocol", Work in Progress, Internet-Draft, draft-yang-dmsc-ioa-task-protocol-02, 2026, <https://datatracker.ietf.org/doc/draft-yang-dmsc-ioa-task-protocol/02/>.

[JSON-LD]

W3C, "JSON-LD 1.1", 2020, <https://www.w3.org/TR/json-ld11/>.

[OWL2-OVERVIEW]

W3C, "OWL 2 Web Ontology Language Overview", 2012, <https://www.w3.org/TR/owl2-overview/>.

[RDF11-CONCEPTS]

W3C, "RDF 1.1 Concepts and Abstract Syntax", 2014, <https://www.w3.org/TR/rdf11-concepts/>.