Research And Implementation Of Fuzzy Clustering Algorithm And Chinese Text Clustering

In recent years, with the development of Internet, various forms of information has cropped up a lot of information for users at the same time, and it adds more difficulty to obtain valuable information for the user . So how can the users access to valuable information that has become an important research topic today. Among the many messages, the amount of information is in text form, therefore, the text dealing has become an important means. Among them, fuzzy clustering algorithm has also become a research hotspot of text processing problems.At present, the classical FCM algorithm is more commonly used as one of fuzzy clustering algorithms, and this algorithm has a better theoretical basis to support, that has been widely used in text processing and fuzzy recognition, etc. However, there are three shortcomings of the classic FCM algorithm: first, the algorithm is more sensitive to the isolation point. Second, the effects of algorithm is good or bad that depends on initial parameter selection. Third, FCM algorithm is not optimal solution and easy to fall into local extreme point. In this paper, the first shortcoming is improved through the membership degree of data weighted. In this paper, the second shortcoming is improved through fuzzy similarity matrix structured in combination with statistical knowledge, in order to achieve a good initial value of selected results. For the third drawback, the introduction of evolutionary strategy optimization based on FCM constitutes a hybrid clustering algorithm, so that such problems can be solved. Through the above research and improvement, this paper presents the new fuzzy clustering NFCM and NES-FCM, and these algorithms have been tested with the classical FCM algorithm. Through the experimental results show that the new fuzzy clustering algorithm NFCM and NES-FCM clustering can improve the accuracy and validity, and give full play its role on local search and global search capability of the FCM algorithm. Finally, a Chinese text clustering model was designed and implemented based on fuzzy clustering of Chinese text clustering and experimental results of the improved fuzzy clustering algorithm show that the improved fuzzy clustering algorithm in the Chinese text clustering has a good performance, also show that the Chinese text of the model is valid.