Towards Causality-Based Conflict Resolution in Answer Set Programs.

Using answer set programming in real-world applications requires that the answer set program is correct and adequately represents knowledge. In this paper, we present strategies to resolve unintended contradictory statements resulting from modelling gaps and other flaws by modifying the program without manipulating the actual conflicting rules (inconsistency-causing rules with complementary head literals). We show how latent conflicts can be detected to prevent further conflicts during the resolution process or after subsequent modifications in the future. The presented approach is another step towards a general framework where professional experts who are not necessarily familiar with ASP can repair existing answer set programs and independently resolve conflicts resulting from contradictory statements in an informative way. In such a framework, conflict resolution strategies allow for generating possible solutions that consist of informative extensions and modifications of the program. In interaction with the professional expert, these solution options can then be used to obtain the solution that represents the underlying knowledge best.