Research of Web information classifying based on neural network

Selecting Web information as important library collections, has become a daily routine of library. Because of the wide variety of Web information, these information need to be processed so that the reader queries easy. Therefore, it is required to study the classification of Web information. In this article, a neural network classifier is proposed, which can classify Web information into each subject. The title and keywords of Web document is choosed as inputs of neural network classifier, then the subjects code of Web document is choosed as outputs of neural network classifier. In order to evaluate the neural network classifier, there are 3000 Web pages for training and 2000 Web pages for testing. Preliminary experiment result shows that the average system performance of our Web information classifier is about 73%.