Based on the Tensor Flow Framework and the Improved Logistic Regression Algorithm, the Detection Model of Weibo Water Army was Constructed

Abstract

With the development of Sina Weibo and the rapid growth of user volume, Sina Weibo has become the largest comprehensive social networking platform. But its hasty progress has further caused the rise of Weibo water army and gradually makes it appear in a rapid increase in the trend. The existence of the water army has a significant impact on the quality of Weibo information, but also causes the shared environment is not healthy and other serious problems. Built on the above reasons, this paper extracts the characteristics of the water are of the characteristics of water army's user attributes, behavioral attributes and temporal attributes by studying the characteristics of the water arm's users and normal users. At the same time, based on the Tensor Flow depth learning framework and the improved logistic regression algorithm, the training model is effectively identified the water army. The accuracy and validity of the model can be demonstrated by the experiment.