Document Type



Many enterprises begin to use Social Network Sites (SNS) as an important channel and platform to do online marketing and reputation management, because users’ interactions in SNS have more effective impacts on customers’ buying decisions and images of enterprises than traditional websites. To do this, the enterprises need to learn and trace users’ sentiments on their products/services for designing appropriate business strategies. In this study, the sentiment community is proposed as a method for this. The sentiment communities with different polarities in SNS usually represent groups of users with different preferences, and discovered sentiment communities is very useful for enterprises to do customer segmentation and target marketing. Also the evolvement of sentiment communities is explored, so that enterprises can easily trace users’ sentiments and learn their diffusions in SNS. In this paper, a novel method is proposed for discovering users’ sentiment communities, and an initial experimental evaluation is executed.