Candidate Set Expansion for Entity and Relation Linking Based on Mutual Entity–Relation Interaction

Entity and relation linking are the core tasks in knowledge base question answering (KBQA). They connect natural language questions with triples base. In most studies, researchers perform these two independently, which ignores interplay between entity linking. To address above problems, some have proposed a framework for joint based on feature multi-attention. this paper, their method, we offer candidate set generation expansion model to improve coverage of correct words ensure that disambiguation objects exist list as much possible. Our first uses initial obtain nodes graph related relation. Second, filtering rule filters out less-relevant candidates expanded set. Third, directly connected added Finally, ranking algorithm An empirical study shows improves recall correctness KBQA. The method entity–relation interaction paper is highly portable scalable. considers connections subgraphs graphs provides new ideas expansion.