A reasoning platform based on the MI Shapley inconsistency value

Sébastien Konieczny (1) , Stéphanie Roussel (1)

1 CRIL - Centre de Recherche en Informatique de Lens

In this paper we show how to build a reasoning platform us- ing an inconsistency value. The idea is to use an inconsistency value for evaluating how much each formula of the belief base is responsible of the inconsistency of the base. Then this evaluation allows us to obtain a strati cation (total pre-order) of the base, that can be used as the preferential input for di erent reasoning tasks, such as inference, belief revision, or conciliation. We show that the obtained operators are interesting and have good logical properties. We use as inconsistency value, the MI Shapley inconsistency value, that is known to have good properties, and that can be computed from minimal inconsistent subsets. We developed a java-based platform, that use the Sat4j library for computing the minimal inconsistent subsets, and that allows to have an e ective way to compute the MI Shapley inconsistent subsets. We implemented also several inference, revision and conciliation methods, that use this inconsistency value. So this provides a complete reasoning platform, that can be used for instance for academic purposes.

Domaines: Intelligence artificielle [cs.AI]

Sébastien Konieczny, Stéphanie Roussel. A reasoning platform based on the MI Shapley inconsistency value. 12th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty (ECSQARU'13), 2013, Utrecht, Netherlands. pp.315-327. ⟨hal-00866813⟩