Abstract: In management information system for EC network resources, in order to display Web pages retrieval results according to different topics to help the users quickly locating the required information and effectively improving user experience, the solution was put forword for related problems. The solution includes user-defined domain dictionary, feature selection based on clustering features dictionary and initial center point construction based on topics domain words. From experimental results, this solution can effectively improve the quality of text clustering for EC Web pages. The text clustering solution can be extended to other specific domains.


Key words: text clustering, k-means, segmentation, feature selection, Web pages retrieval