The candidate will propose and evaluate Web-based multi-agent and formal knowledge models to allow on the one hand the cooperation of distributed AIs addressing different distribution reasons (e.g. computing power, resource availability, legal constraints, etc.) and, on the other hand, the combining of different AI methods (e.g. rule-based, connectionist, evolutionary, etc.). The goal is to perpetually enrich and refine the knowledge collectively maintained by the agents.
The subject includes two challenging research topics:
- Formulate and formalize ontology-oriented knowledge representation of AI data (typically inputs and outputs), metadata (e.g. parameters, provenance, statistics, uncertainty, traces) but also configurations and more complex resources (e.g. schemata, embeddings, mappings) to support AI interoperability and traceability at a knowledge level.
- Design and test a multi-agent model and protocols to orchestrate the interactions between agents embedding different methods of artificial intelligence with the aim of ensuring an optimized collaboration to augment and improve the knowledge shared in the system.