Poster – Paper 649
Forgetting is a operator that eliminates unwanted concepts and roles from a knowledge base, and has found its application in ontology modularity and reuse. Existing forgetting approaches can only handle consistent knowledge bases. We present the first study on forgetting for inconsistent knowledge bases. In particular, we define forgetting via inconsistency-tolerant query answering semantics. Then, we study the computation of forgetting in DL-Lite and provide a graph-based algorithm. Empirical results show that our algorithm is able to scale over large datasets.