Repairing errors in probabilistic databases models using probabilistic abduction reasoning

Abstract

This paper presents a technique to diagnose probabilistic counter examples that are generated when model checking probabilistic databases models against probabilistic properties formulating queries on probabilistic databases. In probabilistic model checking (PMC), a counterexample is a set of paths that satisfies a path formula, whose cumulative probability mass violates the probability bound. The diagnosis is to repair errors in probabilistic PRISM programs of probabilistic databases models using the probabilistic abduction reasoning on independent choice logic (ICL) programs describing the generated probabilistic counterexamples.