An Answer Recommendation Algorithm Based on Semantic Fusion Heterogeneous Information Network

Question answering system is an important direction of natural language processing and has been widely used in various fields. However, existing answer recommendation research mainly confronts the problems of text semantic extraction and multi-feature fusion. In this paper, we propose an Answer Recommendation algorithm based on Semantic Fusion Heterogeneous Information Network, called ARSHIN. Firstly, we employ the PCFG method to mine phrases in texts, and perform phrase embedding to obtain phrase similarity relationships. Secondly, we utilize a heterogeneous information network to fuse text semantics and entity-relationship features. Finally, we give the recommended answers by calculating entity correlations. Comparative experiments on three real datasets demonstrate that ARSHIN has advantages in answer recommendation research.