Retweet Prediction Based on Multidimensional Features

With the wide use of artificial intelligence-driven mobile devices, more and more Chinese people take part in the Twitter-like social sites, which makes Weibo an excellent communication platform. In view of the wide application of information diffusion in various fields, Weibo has become one of the most important research issues in mobile social computing. In Weibo, the retweet statuses of tweets of other users are considered to be the key mechanism for spreading information. How to predict whether a tweet will be retweeted by a user has received increasing attention in recent years. Research shows that the users' retweet behavior is driven by their interest and personality. However, most previous works ignore the roles of users' personality in their retweet behavior. To this end, a prediction model MDF-RP (multidimensional feature-based retweeting prediction) including personality feature is proposed. The prediction model integrates the features from three dimensions, such as author, tweet, and user. And the personality score is obtained based on the well-known Big Five personality trait model. The experimental results under different classifiers show that the performances of MDF-RP features outperform the basic features. And the experiments of cross-validation also demonstrate the stability of MDF-RP features.