Abstract: One of the most important reasons that affect the information retrieval result is the phenomenon of semantic match of user query and the document while syntactic mismatch. Capturing synonym relationships to extend the query term and combining the retrieval result of the simple Bayesian network retrieval model, a compound Bayesian network retrieval model is proposed. The internet topology of the compound model, the retrieval process and the corresponding retrieval algorithms are given. Experimental results show that the model can realize the semantic retrieval, and further optimize the retrieval performance.

Key words: User query, query term, related document, Bayesian network, combine, synonym, information retrieval

摘要： 用户查询与文档之间语义相似或相关，但是词法不匹配是影响信息检索性能的重要原因之一。挖掘术语间的同义词关系，实现查询术语的语义扩展，同时归并简单贝叶斯网络检索模型的检索结果，构造一个复合的贝叶斯网络检索模型。给出复合模型的网络拓扑、检索流程以及相应的检索算法。实验结果表明该模型可以在实现语义检索的基础上，进一步优化检索性能。

关键词: 用户查询, 查询术语, 相关文档, 贝叶斯网络, 归并, 同义词, 信息检索