Mining implict outlier purchasing behaviors from fan group marketing data

Social networks or virtual communities have become new marketing media. Businesses develop new recommender systems by incorporating social network information to promote their products. Because of the characteristic of social network users´ similar tastes, the social recommender systems help the businesses to increase their marketing share easily. In order to attract more new fans, the data mining communities propose methods to analyze purchasing behaviors among social network visitors to improve recommender systems´ efficiency. However, the implicit dissimilar purchasing behaviors among existing fans and the reasons behind such behaviors may also help the business to have a better understanding of community evolution and improve the marketing sale plans to stick the fans by the group. This paper will transform the fans´ purchasing data set into a transaction data set and propose a framework to mine the implicit outlier behaviors with the reasons that induce those behaviors. The experiment is given to show the advantages of the proposed framework.