Evaluation algorithm of false index of social text for opinion leaders

Inhalt:
With the popularity of the Internet, the volume of social media information has increased exponentially, it also provides a channel for the dissemination of false information. The current mainstream researches on text credibility are to evaluate the credibility of user text based on the influence of users themselves, ignoring the impact of user text propagation attributes (such as Like amount) on text credibility, which will result in the not comprehensive social text calculation method of opinion leaders. In order to solve this problem, this paper proposes an evaluation algorithm of social text false index for opinion leaders (FISTOL). Firstly, this paper collects a total of 100 thousand blog datasets and arranges them into the specific topics (Opinion leader dataset, OLD); Secondly, FISTOL calculates the word vector of word by segmentation tool, then convert it into sentence vector through linear combination; Thirdly, this algorithm calculates the false weight values of propagation attributes and the false index of opinion leaders' social text. This paper verifies the influence of the change of each propagation attribute on the false index value and the experimental results show that the false index evaluation algorithm proposed in this paper can correctly reflect the credibility of the text.