基于查询语义树的语义查询扩展研究

本文主要研究了查询语义树的生成策略、用户查询语义的提取机制,以及查询语义树中语义边界的确定方法。通过查询语义树产生候选扩展词,再计算候选扩展词与所有查询项在初检局部文档集合中的共现度,用以评估扩展词质量,使得扩展词与用户查询所蕴涵的主题具有较强的语义相关性。实验结果表明,与传统向量空间模型检索算法比较,查询性能有明显的改善。

The strategy of generating a query semantic tree, the mechanism of extracting user＇ s query semantics from the semantic network of concept, and the method of defining the query semantic boundary in the query semantic tree are studied. After evaluating the quality of expansion terms by calculating their local with original user query terms in the first-retrieved and top-ranked documents, the most appropriate expansion terms are selected. Experimental results show that this approach can improve both precision and recall rate to a great extent compared with the traditional retrieval algorithm of vector space model.