The Untold Story Behind the Recommendation in Micro-blogging Network

Celebrity recommendation is widely-adopted by many micro-blogging services as an important way to enhance users´ experience and the visibility of contents submitted. It is critically important for celebrity recommendation to understand which factors and how these factors affect the probability that the recommended celebrities will be accepted. In this paper, taking the Tecent Weibo as a case, we try to empirically tackle the untold story behind the celebrity recommendation in micro-blogging network. We studied the three potential factors, namely the popularity of the recommended celebrities, the structural similarity and the topical similarity between the recommend celebrity and the target user. As shown by experimental results, the popularity of the celebrity and the structural similarity well reflect whether the recommended celebrity is accepted by the target user, whereas the topical similarity is not a good indicator as commonly expected.