DOI number:

10.4304/jcp.6.11.2357-2364

Affiliation of Author(s):

(1) Dept of Computer Science, Wuhan University, Wuhan, Hubei Province, China

Journal:

Journal of Computers

Abstract:

In Web Community marketing, it is an essential research to discover potential consumer groups based on their interests. In real society, the sociologists deem that the intimate friends show more similarities in their interests. Based on a large number of experimental analysis and data statistics of Web Community users, this paper proves that the view is also suitable for Web Community, and it is used to excogitate to discover the users with similar interests by induction and estimation. This paper compares conversation degree and interest similarity in sample user data, conducts the conjecture experiments and then obtains the relationship between conversation degree and interest similarity by statistics and analysis on the basic conversations. Afterward, we measure the relationship quantitatively and conclude a distribution function. Based on these, an algorithm with regard to mining user groups with similar interest quickly in the web community has been proposed. Finally, we prove the efficiency and accuracy of this algorithm by confidence measurement and algorithm testing. Experimental results show that it is more effective and feasible than other old methods, especially in speed. © 2011 Academy Publisher.

Co-author:

Wei(1),Yu, Zhuo(1), Zhang, Yunlu(1), Shijun(1), Li

Translation or Not:

no

Date of Publication:

2011-01-01