| Summary: | The main objective of this dissertation is to define an argumentation-based negotiation framework for distributed knowledge bases. Knowledge bases are modeling over a multi-agent setting (MAS) such that each agent possibly has an independent or overlapping knowledge base. This dissertation proposes that agents evaluate arguments to obtain a consensus about a common knowledge by both proposing arguments or trying to build opposing arguments against them. Moreover, this proposal deals with incomplete knowledge (i.e. partial arguments) and so a cooperation process grants arguments to achieve knowledge completeness. Therefore, a negotiation of an agent's belief is seen as an argumentation-based process with cooperation; both cooperation and argumentation are seen as interlaced processes. Furthermore, each agent Ag has both set Argue of argumentative agents and set Cooperate of cooperative agents; every agent Ag must reach a consensus on its arguments with agents in Argue, and Ag may ask for arguments from agents in Cooperate to complete its partial arguments. The argumentation-based negotiation proposal allows the modeling a hierarchy of knowledge bases representing, for instance, a business's organization or a taxonomy of some subject, and also an MAS where each agent represents "acquired knowledge" in a different period of time. Furthermore, any agent in a MAS can be queried regarding the truth value of some belief. It depends on from which agent such a belief is inferred, and also what the specification in both Argue and Cooperate is, given the overall agents in the MAS. However, such an answer will always be consistent/paraconsistent with agents' knowledge base involved. This dissertation proposes a (declarative and operational) argumentation semantics for an agent's knowledge base. Furthermore, it proposes a declarative argumentation-based negotiation semantics for a multi-agent setting, which uses most of the definitions from the former semantics.
|
|---|