Topic-sensitive TwitterRank

Basic Idea

• We proposed a topic-sensitive ranking algorithm based on the existing TwitterRank Algorithm.
• Based on the topology level information from TwitterRank, we further utilize the top-specific information, such as the comment preference or the number of tweets on specific topic, namely expertness, to obtain a more precise identification of individual influence.
• A weighted Borda count method is used to combine this two rankings.

Testing

• Testing is performed on real twitter data to compute individual influence, and is combined with individual comments to predict real-time box office prediction.