Text Mining And Its Application In Text Retrieval

Internet is growing explosively and it has evolved into a major source of information in people's ordinary lives. But information on the Internet is distributed, dynamic and diverse which make it difficult to retrieve information on the Internet. How to find the information users needed is a challenge to computer scientists. To alleviate this problem, my thesis has focused researches in the following areas:Firstly, we discussed how to build the indices of text retrieval systems. We focused our discusion on the building, encoding, compressing, storing, retrieving and parallelization of inverted index.Classfication is an effective means of organizing the information on the Internet. Based on the vector space model, we discussed the automatic classification of documents. And we use swarm intelligence algorithm to automatically classfying documents.To help the users to express their query more precisely, we must overcome the vocabulary problem and offer vocabulary-based search aid. In this thesis, we discuss how to use co-occurrence analysis of words and Hopfield neural network to find the relationship between concepts automatically, generating concept space. Using concept space, system can expand users'queries.Finally, we discuss how to implement an effective query interface by integrating the techniques mentioned above.