ESWC 2026 OWL / RDF Ontology CC BY 4.0

AgentO: An Ontology for Modeling Agentic AI Systems

DOI: 10.5281/zenodo.18342624
Venue: ESWC 2026
License: CC BY 4.0
🔗 Ontology  ·  ⚙ GitHub

Bringing Formal Structure to Agentic AI

Agentic AI systems are rapidly deployed as autonomous, goal-directed entities to manage complex, multi-step workflows across domains from cybersecurity and finance to healthcare and scientific research. Yet current implementations remain largely ad-hoc — hard-coded, monolithic, and difficult to reuse or audit.

AgentO is an OWL/RDF-based ontology that formally represents the core concepts, components, and interactions underpinning agentic AI workflows. It provides a standardized vocabulary for modeling agents, tasks, workflows, tools, goals, and resource dependencies — enabling modular design, cross-framework reusability, automated reasoning, and system auditing. Accompanying the ontology is a knowledge graph instantiating 66 agentic workflow patterns extracted from four major frameworks.

Core Concepts in AgentO

AgentO builds upon PROV-O, P-Plan, and BEAM and defines the following classes and relationships. Each concept is formally represented as an OWL class enriched with data and object properties.

:TeamCoordinated group of agents (beam:System)
:LLMAgentLanguage-model-backed autonomous agent
:HumanAgentHuman-in-the-loop participant
:TaskActivity contributing to a goal (pplan:Step)
:GoalDesired state for an agent or team
:ObjectiveCollective objective for a team
:ToolInstrument extending agent capability
:ResourceAsset consumed or produced in tasks
:WorkflowPatternReusable structured template (pplan:Plan)
:WorkflowStepIndividual phase within a workflow
:CapabilityAbility an agent can perform
:PromptStructured instruction for an agent
:MemoryPast information for reasoning support
:KnowledgeBaseStructured info an agent references
:LanguageModelUnderlying model of an LLM agent
:ConfigRuntime / design-time configuration
:EnvironmentOperational context of an agent
:ConstraintRule restricting agent decisions

Frameworks & Patterns Collected

AgentO was built and evaluated using 66 agentic workflow templates drawn from four popular open-source frameworks.

66Patterns Total
4Frameworks
24Manually Verified
27Issues Resolved
Framework Language Patterns Repo
AutoGen Python 6 ksm26/AI-Agentic-Design-Patterns-with-AutoGen
CrewAI Python 16 crewAIInc/crewAI-examples
LangGraph JavaScript 9 langchain-ai/langgraphjs-gen-ui-examples
Mastra AI TypeScript 35 mastra-ai/mastra

KG Construction & Access

The knowledge graph was constructed using an automated LLM-driven pipeline that parses raw agentic framework code, maps concepts to AgentO via prompt-guided extraction, and emits RDF Turtle. All extractions used gpt-5-mini via API, processing ~2.8M input tokens and ~570K output tokens at a total API cost of $2.72.

The resulting graph is publicly hosted in a triple store and accessible via a SPARQL endpoint for downstream use cases including agent composition, design inspection, and explainability reasoning.

2.8MInput Tokens
570KOutput Tokens
$2.72Extraction Cost
~108sAvg. per Pattern

Resources

Authors

Andreas Ekelhart
University of Vienna
SBA Research, Vienna
aekelhart@sba-research.org
Kabul Kurniawan ✉
Universitas Gadjah Mada
Yogyakarta, Indonesia
kabul.kurniawan@ugm.ac.id
Fajar J. Ekaputra
Vienna University of
Economics and Business
fajar.ekaputra@wu.ac.at
Elmar Kiesling
Vienna University of
Economics and Business
elmar.kiesling@wu.ac.at

Funded by LPDP (Indonesia), Austrian Science Fund (FWF), Austrian Research Promotion Agency (FFG), and SBA Research COMET Center.

How to Cite

If you use AgentO or the accompanying knowledge graph in your research, please cite the following:

@inproceedings{ekelhart2026agento,
  title = {AgentO: An Ontology for Modeling Agentic AI Systems},
  author = {Ekelhart, Andreas and Kurniawan, Kabul and Ekaputra, Fajar J. and Kiesling, Elmar},
  booktitle = {Extended Semantic Web Conference (ESWC 2026)},
  year = {2026},
  doi = {10.5281/zenodo.18342624},
  url = {https://agentic-patterns.github.io},
  license = {CC BY 4.0 International}
}