Probabilistic Belief Embedding for Knowledge Base Completion

This paper contributes a novel embedding model which measures the probability of each candidate belief 〈h, r, t,m〉 in a large-scale knowledge repository via simultaneously learning distributed representations for entities (h and t), relations (r), and even the words in relation mentions (m). It facilitates knowledge completion by means of simple vector operations to discover new beliefs. Given an imperfect belief, we can not only infer the missing entities, predict the unknown relations, but also tell the plausibility of that belief, just by exploiting the learnt embeddings of available evidence. To demonstrate the scalability and the effectiveness of our model, we conduct experiments on several large-scale repositories which contain hundreds of thousands of beliefs from WordNet, Freebase and NELL, and compare the results of a number of tasks, entity inference, relation prediction and triplet classification, with cutting-edge approaches. Extensive experimental results show that the proposed model outperforms other state-of-the-art methods, with significant improvements identified.