基于用户日志挖掘的查询扩展方法

A NEW QUERY EXPANSION METHOD BASED ON USER LOGS MINING

查询扩展可以有效地消除查询歧义,提高信息检索的准确率和召回率。通过挖掘用户日志中查询词和相关文档的连接关系,构造关联查询,并在此基础上提出一种从关联查询中提取查询扩展词的查询扩展方法。同时,还提出一种查询歧义的判别方法,该方法可以对查询词所表达的检索意图的模糊程度进行有效度量,也可以对查询词的检索性能进行预先估计。通过对查询歧义的度量来动态调整扩展词的长度,提高查询扩展模型的灵活性和适应能力。

Query expansion has long been suggested as an effective way to eliminate query ambiguity and improve the precision and recall rates of information retrieval.By mining relations among phrases and relative documents in user logs,the paper constructs related queries,based on which proposes a query expansion method to extract query expansion words from related queries.Meanwhile the paper introduces a method to measure the ambiguity of user queries,which can calculate the fuzzy degree of uses’ search intentions and can estimate the performance of search sessions in advance.The query ambiguity measurement helps dynamically adjusting the number of expanded terms,so that the flexibility and adaptability of query expansion model is improved.