信息检索结果隐式多样化排序方法研究

Research and Analysis on the Implicit Diversification Ranking Method for Information Retrieval

针对信息检索隐式多样化算法展开研究。介绍了检索结果重排序中的最大边际相关度(MMR)算法、相对熵(KL)算法、现代投资组合理论(MPT)3个隐式多样化算法,采用此3个算法对所选的已排序的文档数据集进行重排,通过对排序结果进行评价来对比三者的性能。得出隐式多样化方法中当相关性和多样性以一定比例的线性组合时,会使最终的检索结果在一些评价指标上相对于原始结果有所提高。

Research on implicit diversification algorithm for information retrieval. On this basis, the documents are diversified. In this paper, first we introduce the maximum marginal relevance （MMR） method, Kullback - Leibler divergence （KL） method and modem portfolio theory （ MPT）, then use these three methods to re - rank data sets, and then evaluate and compare the performance of the three methods. Finally, it is concluded that the implicit diversification methods can improve the performance compare with the original results when the correlation and diversity in a certain percentage of linear combination.