Diffusion of recommendation through a trust network

Several attempts have been made to analyze information diffusion on online knowledge sharing sites. In this poster, we report a preliminary study analyzing @cosme, a viral marketing site, which is among the largest community sites in Japan. The notable characteristics of @cosme are that users can bookmark their favorite reviewers; moreover, they can post their own reviews of products. In other words, a user puts trust in some other users for their reviews. This trust relation affects the opinions of users: A user refers to the ratings and opinions from the user's own favorite users when purchasing products; some users come to be trusted by a user based on mutual similarities of ratings and opinions. This bidirectional interaction between trust and opinion is an important phenomenon that elucidates consumer behaviors in online communities. In this paper, we describe an overview of the data in @cosme, analyses of recommendation through the trust relation, and concept for future analyses.