Research of Texts Automatic Categorization Algorithm Based on Weighted Fuzzy Reasoning Network

Automatic categorization of texts is one of the most important programs in information management fields. Text categorization method based on weighted fuzzy reasoning network is put forward according to the text characteristics, fuzzy in categorization properties, and the importance emerged in characteristics information against text class. Taking text categorization training in the network as a linear program, the related learning algorithm is conducted by the minimum mean-square value principle and idea of iteration. Finally, analyze data about text information of some traveling Web pages to verify the validness of this algorithm. At the same time, it is showed that the problem of local minimum usually caused by traditional grad decline algorithm is avoided because of no grad calculations during iteration.