信息检索中支持隐式时间查询的文档排名方法

Ranking methods for implicit temporal queries in information retrieval

针对时间信息对检索结果的影响,提出一种支持隐式时间查询的文档排名方法。结合DBpedia这种特殊的语义网以及检索结果中排名较前的文档对排名结果的影响,分析用户查询的时间意图,利用排名模型计算文档时间相关性方面的得分,通过线性结合方法综合考虑时间相关性和内容相关性,对检索结果重新排序。实验结果表明,该算法能够在检索流程中有效地分析查询的时间意图,在排名模型中引入时间因素有利于提升检索结果性能。

Aiming at the influence of temporal information on the retrieval results,a document ranking method for supporting implicit temporal queries was proposed.DBpedia knowledge base and top k documents of results that only considered content relevance were used to analyze temporal intent of queries.Ranking models were used to calculate temporal relevance score.Documents were re-ranked according to the linear combination of temporal relevance score and content relevance score.Experimental results show that the proposed method does well in analyzing temporal intention of queries,and retrieval performance is improved after adding time factor to the ranking model.