基于深层特征和集成分类器的微博谣言检测研究

Research on detecting micro-blog rumors based on deep features and ensemble classifier

微博中存在着大量的虚假信息甚至谣言,微博谣言的广泛传播影响社会稳定,损害个人和国家利益。为有效检测微博谣言,提出了一种基于深层特征和集成分类器的微博谣言检测方法。首先对微博情感倾向性、微博传播过程和微博用户历史信息进行特征提取得到深层分类特征,然后利用分类特征训练集成分类器;最后利用集成分类器对微博谣言进行检测。实验结果表明,提出的基于深层特征和集成分类器的方法能够有效提高微博谣言检测的性能。

There are a large number of false information and rumors in micro-biog. The wide-spread of rumors have seriously affected the social stability and damaged individual and national interests. In order to detect the micro-blog rumors effectively, this paper proposed a method based on deep features and ensemble classifier. Firstly, it extracted features from the sentiment orientation, propagation and user＇ s historical information of micro-biog. Then it trained the classifier by using these deep fea- tures. Finally, it detected micro-blog rumors based on the ensemble classifier. The experimental results show that the proposed method can effectively improve the classification performance.