Application Of Latent Semantic Indexingin Chinese Information Retrieve

In today's world, we are facing to the vast information on the internet. How to quickly and accurately retrieve the required content of the people has become an great challenge. At present, commonly used keyword search or vector space model are the existence of certain defects, such as unable to deal with synonyms.Latent Semantic Indexing(LSI) has a remarkable performance when it treat at synonyms. And by selecting the appropriate space dimension to compressed the Latent Semantic space, we can effectively improve the accuracy of search results and retrieval speed.This article will focus on LSI to make a deeply discussion. We discuss the principle of LSI, Related technologies and the mathematic principle of LSI. Through the examples to introduce how to use the LSI model for Chinese text processing. And for problem of Latent semantic space dimension(k value) selection, we proposes a method based on user feedback, and through the experiment compared the results which before feedback and after.Finally we use latent semantic indexing module to be experimental platform, and analysis the results of the improved LSI methods, At last, we get the final conclusion.