Agent-based theorem proving

Линиите свързват координатора с партньорите.

Накратко на български

Цел на проекта

In this project we transfer and apply multi-agent techniques to automated theorem proving. The goal is to achieve more flexible organization of heterogeneous reasoning techniques within a theorem-prover to provide a better approximation of human reasoning and to solve more complex problems. For this we will investigate several topics. -Distributed Proof Search: We have to examine possible dependencies between parallel proof attempts I how backtracking can be organized and how a final proof object can be reconstructed. -Resources: Appropriate resource concepts have to be identified to keep proof search tractable. -Flexible Cooperation: This is achieved by dynamic formation of clusters of reasoning agents to either cooperatively or concurrently solve problems. The concise communication of partial proofs between agents is essential to keep communication time low. -Architecture for Reasoning Agents : The reasoning agents need to be embedded into an agent architecture that allows them to recognize their own abilities and those of other agents in order to negotiate about forming clusters and resource allocation. After modeling an appropriate system architecture we plan to implement the working system and to test our concepts with a case study to experiment both with the theorem proving aspects as well as the multi-agent aspects of the system.

Оригинален текст от CORDIS (на английски).

Участници

• UNIVERSITY OF BIRMINGHAM · BIRMINGHAMКоординаторОбединеното кралство

Връзки

• Виж в CORDIS