A further step for efficient corrections of inconsistent probabilistic data sets

Partial conditional probability assessments are having renewed attention and one of the more compelling need associated with them is the of merging several sources of information. We focus here on the consequent mandatory task of correcting inconsistent probabilistic databases. Since probabilistic satisfiability problems (PSAT) has mainly suffered of space complexity in their original formulations, we propose an efficient method for correcting incoherent (i.e. inconsistent) conditional probability assessments. This method is based on L1 distance minimization and Mixed Integer Programming (MIP) procedures, taking into the right consideration the compulsory need and benefits of dealing with different “zero layers”. Through a simple prototypical example, we illustrate the feasibility and the peculiarities of the proposed procedure.