Twitter is emerging social media. There is an important practical significance to study its communication effect in many aspects of the Marketing Management, Public Sentiment, Hot Extraction and so on. The retweeting scale is an important indicator to reflect the communication effect. Based on analyzing the factors, a prediction algorithm based on SVM was proposed for retweeting with five features: publisher influence, acceptor activity, interest level, content importance and intimacy. Furthermore, the prediction algorithm for retweeting scale was proposed based above, the method to evaluate the prediction accuracy was given. At last the experiment with Twitter data showed a good result that the prediction accuracy of retweeting scale was up to 86.63%.
I. INTRODUCTIONWITTER is a platform based on relationship to share, propagate and receive information. People can set up individual communities through Web, mobile phone and other client, share immediately with about 140 words of text messages. Its originality, timeliness, grass-roots, arbitrariness and other characteristics brings a new type of social network to the Internet. The number of registered users has grown up to 200 million since 2006. As emerging social media, Twitter has become not only a tool to express opinions and communicate with others, but also a method of information release, public relation and marketing for government, enterprises, and organizations.Fundamentally speaking, Twitter is still communication media, and the ultimate goal is to deliver message with the best communication effect. As emerging social media, there are many unique properties compared with traditional media. Therefore, it's essential to study how to disseminate information effectively and efficiently in the new media environment of Twitter. Communication effect is a concept of the communications [1]; it refers to the overall impact and result on the acceptor and society from dissemination activities of newspaper, radio, television and other mass media. The communication effect is a abstract concept without recognized uniform evaluation criterion. Different media have different indicators to evaluate their effects, for example, circulation for newspaper, audience rating for television program, box office for movie and so on. The
