Abstract: Aiming at the problem that in the field of natural language processing,  incomplete knowledge graphs led to the entity association expansion, which required additional inference and reasoning to make the derivation process of answers  more complex, we proposed  a knowledge graph question answering algorithm  RMP-KGQA that combined relational memory and  path information. The algorithm used a relational memory network to solve  the problem of inconsistency between the problem and the knowledge graph mapping space, and  enriched the scoring function with its path information, significantly enhancing the accuracy and robustness of the intelligent question answering retrieval system. The experimental results show that on the WebQSP and WebQSP-50 benchmark datasets, the accuracy of RMP-KGQA  increases by 2.8 and 2.4 percentage points respectively compared to EmbedKGQA. Ablation experiments further verify  the key roles of relational memory perception and path information in the model. Therefore,  RMP-KGQA is an effective method for solving  multi-hop knowledge graph question answering problems  in complex environments.

Key words: knowledge graph question answering, knowledge graph, knowledge graph embedding, relational memory network