Research on Sentiment Analysis of Micro-blog's Topic Based on TextRank's Abstract
Kong Ying, Pan Jingchang, Wu Minglei · 2017
The widespread rise of social media makes the social network more complex, and the development of Web2.0 technology makes the users' online participation greatly increased. According to the psychological research, the groups are easier to establish extreme positions than members that is what is called the group polarization phenomenon. If we do not control and guide the network rumors, the network group events and even the violences are easy to occur. In order to stop the unexpected event, the positive and negative events should be discerned in time. Due to the widespread use of micro-blog platform, the speed and breadth of propagation of the micro-blog hot topics is not expected, so it is important to distinguish the sentiment properties of topics. In order to solve this problem, the spam filtering mechanism based on TextRank is proposed, and the sentiment of the micro-blog hot topics after filtration is analyzed. Compared to the classifier based on sentiment dictionary and the classifier based on Naive Bayesian, the proposed algorithm improves the accuracy of sentiment analysis and recall rate effectively.