The Semantic Query Expansion Model Based On Latent Semantics Index Model

The information technology has been entered into each corners of our society, With the rapidly increasing of the data in the internet, It's not enough for us to collect and manage the information only depend on our brain. For searching the information efficiently, It's very import to develop a method that can search document automaticly. At present all the mature information retrieval model use the index words as query and document , which searching document through the key words. The latent semantic index is a new information retrieval model which has been developed during the last ten years, LSI make full use of the latent semantic information in the document .It can overcome the synonymy and polysemy problem caused by the traditional key word matching model. LSI is more efficiently than the other traditional IR models, It can improve the precision of IR system from 20% to 30%.The Object of this paper is to improve the information retrieval system performance based on the LSI model. In this paper , We use the co-occurrence path to explain the relationship between the index words and extract the semantic information in the term-term matrix to expand the query. Then we propose a new information retrieval named "semantic query expansion model", This model casts off the complex syntax analysis and overcome the weakness of the old IR system, It can improve the performance of the information retrieval system.Moreover, we design and realize the new model by java programming language, We compare the information retrieval performance between the semantic query expansion model with the traditional LSI model. At last from the result of the comparing experiment we prove the new model is more efficiency than the old LSI model. So we can conclude that the semantic query expansion model is a more efficiency query expansion information retrieval model.