EXPANSION OPERATORS FOR MODELLING AGENT REASONING IN POSSIBILISTIC DEFEASIBLE LOGIC PROGRAMMING

CARLOS IVÁN CHESÑEVAR; GUILLERMO SIMARI,; LLUÍS GODO,; TERESA ALSINET,

Bruselas

Conferencia; European Conference on Multiagent Systems (EUMAS); 2005

AgentLink

Possibilistic Defeasible Logic Programming (Pdelp)is a logic programming language which combines featuresfrom argumentation theory and logic programming,incorporatingas well the treatment of possibilistic uncertainty andfuzzy knowledge at object-language level.Defeasible argumentation in general and Pdelp in particularprovide a way of modelling non-monotonic inference.When modelling intelligent agents, capturingdefeasible inference relationships for modelling argumentand warrant is particularly important, as well as thestudy of their logical properties.This paper analyzes two specialized non-monotonicoperators for Pdelp which model the emph{expansion} of a givenprogram progPby adding new weighed facts associated with argument conclusions andwarranted literals. Different logical properties are studied and analyzed, providing useful comparison criteria that can be extended andapplied to other argumentation frameworks.