You are here

Reconciling Agent Theories of Interaction Using Argumentation

SSP Seminar Series
Paul Martin

Knowledge dissemination is motivated by and in turn motivates interaction within theory-driven multi-agent systems. Agents form theories with which they interpret their environment, drawing upon accessible data sources which may be incomplete or themselves the product of interpretation.

Considered here is a possible framework in which agents advance arguments deciding the course of an interaction (specified using LCC) as those interactions are enacted, determining a joint theory for an interaction (called here a portrayal) which in turn motivates further interaction and the revision of agent theories. Explored here is how principles from argumentation can be used to resolve conflicts between agent beliefs and how over the course of multiple interactions, a kind of targeted joint consistency between disparate knowledge sources may be attainable, providing heterogeneous agents with a broader basis for collaboration.

Date and time: 
Tuesday, 10 June, 2008 - 12:00
60 minutes