Design And Implementation Of Network Automatic Text Classifier

With the rapid development of Internet technology, there are abundant, semi-structed, isomeric and dynamic information resources on Internet . Among these information resources, most of all exist in the from of text. In order to be easy to from the Chinese of magnanimity excavate out the valuable information in the civilian shelves, active organisation and classification must be carried on to the mass character shelves collected.On basis of analyzing the present research and existing question of text mining, this thesis mainly studies a text classifier based on neural networks, such as feature extraction , dimension reduction, hierarchical classification and classifier training, are discussed in details. The main research works are shown as follows.(1)Introduce the basic theory and the relevant knowledge of Data mining and text mining,and analyze the research background, the present situation and the existing questions of text mining and text classification.(2)Analyze the essential technologies detailedly in the process of text classification, such as text pre-process, participle technology, text expression, weight computation, feature selection, hierarchical classification and extraction, dimension descending technology.(3)Propose one kind of text classifier based on neural net works.Using artificial neural networks as text classifier, dimentsion of vector can be reduced in the process of text classification, and imports Latent Semantic Indexing for dimension reduction.So the classfied speed and system performance are improved.