Research And Application Of Combinatorial Clustering Methods In The Text Clustering

With the widely use of Internet , various types of text messages continue to growth explosively in the networks, their number is incalculable ,It is difficult for people to identify the related topics from the enormous text messages, Using computer to help to the classification is a good method, So text clustering has become a very hot research topic in data mining research areas.There are a lot of clustering algorithms now ,which mainly concentrate in singlely clustering and their related parameters' improvements .This paper focuses on combinatorial clustering methods.First of all, three kinds of popular text clustering algorithms (SOM clustering algorithm, K-means clustering algorithm, FCM clustering algorithm) have been introduced detailly, their advantages and disadvantages also be analysised.Then, combined with the methods of the feature selection ,two clustering flow models are proposed , Their effectiveness and characteristics have been explained theoretically ,and then the corresponding clustering algorithm (DSOM-FS-K-means algorithm and DSOM-FS-FCM algorithm) have been described in detail..Finally, in order to verify the effectiveness of combinatorial clustering algorithms, we compare these two algorithms to their corresponding single clustering algorithms or their corresponding Composite clustering algorithms without feature selection, the results analyze of the experiments proved that the combinatorial clustering algorithms are superior to the corresponding single clustering algorithms.