Deliberative DeLP agents with multiple informants

Abstract

In this paper we define a trust-based argumentative reasoning formalism where the source of the received information is used to decide the warranted conclusions. In the proposed formalism, the agent’s tentative conclusions are supported by arguments, and these conclusions can in turn be attacked by other arguments, referred to as counter-arguments. The inference mechanism compares arguments and counter-arguments to decide which conclusion prevails. We propose a novel development of different argument comparison criteria based on trust. Thus, given a particular application domain, the more appropriate criterion can be used. Finally, since a realistic scenario should consider that trust on informants is usually subject to change, a complete change theory over the agents’ trust is introduced.