基于关联规则挖掘的查询扩展检索性能研究

介绍了基于关联规则的局部反馈查询扩展基本思想，重点研究关联规则支持度、置信度和扩展词数量对查询扩展检索性能的影响。实验结果表明，这种查询扩展的检索性能对其支持度、置信度以及扩展词数量比较敏感；从关联规则获得的扩展词可以分为两类，即与原查询正相关的扩展词和与原查询负相关或者假相关的扩展词（即噪音），前者可以提高和改善查询扩展的检索性能，而后者只能降低其检索性能。

Studies on Retrieval Performance of Query Expansion Based on Association Rules Mining

The basic idea of the query expansion of local feedback based on association rules is given.And then the impact of support and confidence of association rules, and the number of expansion terms on its retrieval performance are researched principally.Experimental results indicate that the its retrieval performance is sensitive to support and confidence of association rules,and the number of expansion terms,and expansion terms from association rules can be classified into two categories: expansion terms that positive correlation of original query and expansion terms that fake or negative correlation related to original query(namely noises),the former can improve retrieval performance of query expansion,while the later makes its retrieval performance to fall.