Abstract
Computing of conflicting elements in prioritized \dllite{} knowledge bases when the assertions are provided by multiple and conflicting sources is an important task to repair these kinds of knowledge bases. For this purpose, several algorithms have been proposed in the literature for computing one minimal conflicts set in \dllite{} knowledge base. Even though, the proposed algorithms for repairing \dllite{} knowledge bases have proven their effectiveness, they are still lacking in efficiency. In this paper, we demonstrate how GPUs can accelerate the selection of a set assertional base conflicting (contradictory elements), leading to further reduction in runtime. This way of programming uses data parallelism to minimize execution time compared to sequential programming. Our experimental studies demonstrate the potential to achieve a speed-up of up to 12.68\(\times\) by employing modern GPUs for computing conflicting sets under inconsistency in lightweight knowledge bases.