Abstract
In recent years, information analysis is an important issue, especially on the Internet. With the huge and changeable information flood on the web, web text mining can help us to mine useful information effectively and efficiently. Web text mining includes text categorization, text cluster, association analysis, and trend prediction. In this study, we utilize a revised Fuzzy c-Means algorithm of the data mining techniques to text categorization and text clustering. However, the existing algorithm requires providing the number of clusters a priori, which is not practical. Therefore, in this study we intend to solve this uncertain problem by fuzzifying the total degree of belonging to become . This leads to the Type-2 fuzzy numbers in the degrees of belonging and the model becomes a -Fuzzy Means for differentiation. It has been shown that with such relaxation, text mining is more flexible in applications. Meanwhile, we have designed a classifier with our developed ranking method so that the shortcomings of the existing methods can be overcome. A case of mobile-phone information provider has been used to demonstrate the proposed method and the results are comparatively satisfactory.