Set-based Noise Elimination for Is-a Relations in a Large-Scale Lexical Taxonomy

As the significance of knowledge base has been widely accepted during the past decade, how to efficiently eliminate the noises in the knowledge base becomes a key problem since the automatically constructed knowledge base usually contains lots of noises that disturbs its application. Based on the observation for Is-a relations that the real entities of a concept A always share several same ancestors besides A, we come up with an Is-a relations noise elimination approach. In this paper, we will elaborate on this approach and explain the pseudocode of it. Our experimental results demonstrate that such an approach is capable of eliminating the noises in the knowledge base efficiently.