Retweeting is a powerful driving force of information propagation on microblogging. How to identify effective retweeters of a message (called "key retweeter prediciton" problem) has then become a significant research topic. Conventional approaches addressed this topic mainly from two aspects by analyzing either the personal attributes of microblogging users or the network structure of user graphs. However, according to sociological findings, the author-retweeter dependence also plays a crucial role in the influence of propagation. In this paper, we proposed a novel model for key retweeter prediction problem by incorporating the auxiliary relations between a message author and potential retweeters. Without loss of generality, we formulate the relations from four aspects, including status relation, temporal relation, locational relation, and interactive relation. In addition, we propose a novel method called "Relation-based Learning to Rank (RL2R)" to determine the key retweeters for a given tweet, by ranking potential retweeters in terms of the spreadability. Experimental results show that our method outperforms the state-of-the-art algorithms on top-k retweeter prediction, with a significant 19.7%-29.4% averagely relative improvement. The findings give new insights into the understanding of user behaviors on social media for key retweeter prediction.
|Number of pages||13|
|Journal||IEEE Transactions on Knowledge and Data Engineering|
|State||Published - 1 Mar 2020|
- Information propagation
- Key retweeter prediction
- User behavior