The project can be applied to debugging (diagnosis) of SAT (SAT4J), Constraints (Choco, JSolver) and OWL (any OWLAPI compatible reasoner) knowledge bases as well as ASP programs. At the moment the project team focuses on development on the OWL debugging module (see Publications for details). Therefore, CP and SAT modules might be slightly outdated.

Given user requirements such as consistency, coherency of ontology and test cases and a background knowledge, which are formulas assumed to be correct, the library computes:

• a set of conflicts, which are irreducible subsets of KB formulas causing violation of requirements/test cases;
• a set of diagnoses, which are minimal subsets of KB that should be changed in order to satisfy all the requirements/test cases.

In the standard implementation conflicts are computed using Junker's QuickXPlain algorithm whereas diagnoses by a corrected version of the HS-Tree algorithm presented in Reiter's "A Theory of Diagnosis from First Principles".

Note that the test cases can be specified as logical formulas that must be entailed/not entailed by the KB or some other functions/metrics that allow to decide whether KB corresponds to the target one that the user intended to formulate.

In general case the library can return many diagnoses as there may exists multiple ways to repair a KB. Therefore we developed an interactive debugging algorithm that allows a user to reduce the number of diagnoses by answering a number of queries, i.e. whether some set of formulas should be entailed by the target KB or not.