Artefacts
Collaborative Discussion
Summary Post
The discussion around ontology languages for software agents on the web has reinforced and refined my initial argument that OWL 2 (Web Ontology Language) and JSON‑LD (JavaScript Object Notation for Linked Data) are the most relevant technologies in the emerging landscape of agentic AI. Our recent module units emphasise the importance of formal semantics, logical consistency, and machine‑interpretable knowledge structures. These principles align strongly with the roles these two languages play.
My peers broadly agreed with the central claim that OWL 2 remains essential due to its grounding in Description Logic and its capacity for automated reasoning. Fatema and Jordan both highlighted, that OWL 2's value is not diminished by the rise of LLMs, but rather amplified. Jordan's reference to iterative refinement loops using reasoners such as HermiT and Pellet supports my view that OWL 2 can act as a stabilising symbolic layer for probabilistic models (Magana and Monti, 2025). This reinforces the neuro‑symbolic perspective I initially pointed to and strengthens the argument that formal ontologies are indispensable for ensuring semantic coherence and explainability.
Both Fatema and Matthew raised an important distinction between ontology modelling and data serialisation, arguing that JSON‑LD is not itself an ontology language but a transport and interoperability layer built on RDF. Their feedback sheds light on a gap in my understanding on the difference between ontology and other forms of knowledge representation, and where the distinction lies, which is a gap that I look to bridge in the future. However, both colleagues agreed that JSON‑LD's power lies in enabling decentralised agent discovery, capability description, and cross‑platform communication, with their preserved comment stating that JSON-LD does not provide deep semantic expressiveness.
Jordan's contribution further situates OWL 2 and JSON‑LD within the broader shift toward a "web of agents" (Petrova et al., 2025). I agree that their complementarity, formal semantics plus flexible exchange, embodies the hybrid architectures now defining agentic AI.
Overall, the peer feedback has strengthened my position and clarified the layered roles these technologies play in ensuring both epistemic reliability and scalable interoperability in modern AI ecosystems.
References
Magana, I. and Monti, M. (2025) Enhancing Large Language Models through Neuro-Symbolic Integration and Ontological Reasoning, arXiv.org. Available at: https://arxiv.org/abs/2504.07640v1 (Accessed: 23 December 2025).
Petrova, T. et al. (2025) From Semantic Web and MAS to Agentic AI: A Unified Narrative of the Web of Agents, arXiv.org. Available at: https://arxiv.org/abs/2507.10644 (Accessed: 23 December 2025).