基于Markov网络的信息检索扩展模型

Extended information retrieval model based on Markov network

为了解决信息检索性能较差的问题,查询扩展将索引项之间的关系以及文档之间的相似度引入到检索中,这个过程可以通过构造知识网络来进行。M arkov网络是一种有效的知识关联图形表示方法,可以从实例数据训练获得。本研究提出并实现了基于M arkov网络的信息检索扩展模型,通过对文档集的学习,构造了关于索引项和文档的M arkov网络,将有利于检索的信息加入到检索中。实验表明,基于M arkov网络的信息检索扩展模型优于BM 25模型。

To improve information retrieval performance, the query extension incorporates similarity between terms documents into information retrieval by constructing a knowledge network. The Markov network is an effective graphical model of representing the relevance among knowledge and can be trained from real data. An extended information retrieval model based on Markov network was proposed and implemented in this paper. Through term and document networks trained from corpus, a Markov model of terms documents was ...