A domain entity answer learning ranking method of integrating multi-feature

作者：Meng Zhipeng; Yu Zhengtao^*; Su Lei; Guo Jianyi; Zong Huanyun

来源：Journal of Computational Information Systems, 2012, 8(3): 1139-1147.

• Domain entity
• Logistic regression
• Multi-feature
• Ranking learning
• Ranking model

In view of characteristics of the factoid question and the list question of the Question Answering System (QA), this paper proposed a domain entity answer ranking model which integrates multiple features. First, for the specific domain, we use conditional random fields to identify entities; then retrieval the candidate answers documents that related to the query, take the candidate answer documents as the bridge of the connecting the topics of the question and the answer entity, use logistic regression algorithm to build the domain entity answer ranking model with various features of domain candidate answer entity and documents and the relevance of the answers, get the relevance of the answer entity and the answer question through integrating the relevance of candidate answer documents and the topics of the query, and then rank according to the relevance. Finally, we carried on the tourism domain entity answer ranking experiment, and the results show that the precision and recall rate of the answer ranking have been improved obviously compared to other methods.

• 出版日期2012