Research on privacy disclosure detection in microblog

In order to detect privacy which could by any possibility disclose in microblog and avoid to harm users, this paper proposes a privacy disclosure detection method for microblog. In this method, it implements personalized privacy disclosure detection by combining detecting privacy disclosure in microblog content and computing privacy concerns of users. In fact, user privacy concerns have a significant impact on risk warning on privacy disclosure. Once privacy disclosure occurs, users with high privacy concerns should be given a serious risk warnings. To implement this method, firstly, it uses convolutional neural network to classify the categories of privacy and non-privacy for microblog, and then applies the support vector machine method to detect privacy in the category of privacy microblog. Meanwhile, it uses decision tree method to determine the privacy concerns of users. The comparison experiment results show that the accuracy of privacy disclosure detection method proposed in this paper is to 91. 6% which have obvious effect.