A prediction model of micro-blog affective hotspots based on SVM collaborative filtering recommendation model.

As innumerable events are reported on Weibo every day, it becomes especially important to predicate the event development trends early as possible. Weibo has been indispensable to the public life; its topic heat predication has become one of the hotspot subjects in data excavation for it provides the basis for public opinion monitoring. In this paper, a non-parametric method is adopted to predicate the heat variation of topic. This method, while maintaining the lower error rate, can effectively predicate whether the topic is heating up. Through the experiment, the parameters can be set flexibly to balance the detection time, true positive rate and false positive rate. The algorithm proposed in this research is effective and extendible, it is believed to help the Sina weibo developer and government in public opinion monitoring. According to this requirement, a non-parametric method is proposed to predicate the development trend of Sina weibo topics. [ABSTRACT FROM AUTHOR]