Arguments Evaluation in Multi-Agent Systems

Abstract

In this paper, a novel problem within the field of Epistemic Logic and Justification Logic is presented. This problem can be stated as follows: how epistemic agents should prefer one justification to another for supporting a given proposition in order to have better pieces of evidence. Unfortunately, it seems that preference between justifications is a context-dependent notion and hence it has to be analysed attending to the context. Consequently, we focus on the analysis of a particular kind of justifications, deductive arguments, and we make some other contextual assumptions. This movement brings us directly to the problem of argu- ment evaluation. According to our analysis, the preference between deductive arguments is the result of a process of argument evaluation in which the agent makes her arguments go through a test consisting on several filters or criteria. As a consequence of this approach, we have that the notion of preference between deductive arguments can be reduced to a list of different epistemic and argumentative notions. For modelling our analysis, different formal tools taken from Epistemic Logic, Justifi- cation Logic, Logics for Belief Dependence and Preference Logic are used. Due to space limitations, we just present here an axiomatic system that can be proven to be sound and complete with respect to its semantics.