| With the rapid development of social media,Sina Weibo,Wechat and other social media platforms have become the main channels for people to obtain,share and disseminate information.However,social media platforms has a low threshold for creating and disseminating information,which brings convenience and richness to people’s lives,and at the same time provides a powerful channel for the dissemination of disinformation.The dissemination of disinformation has brought great negative effects to the society.Therefore,how to effectively identify disinformation has attracted widespread attention from related researchers.At present,the field of disinformation recognition is not enough to mine sentiment context features.To a certain extent,it ignores the connection between contexts or the hierarchical relationship between sentences and documents.This research focuses on using contextual sentiment features to explore the role of implicit features in identifying disinformation from a sentiment perspective,accurately extracting sentiment features in information texts and user comments,and assisting in identifying the true and false of auxiliary identification information.The main research work of this paper includes:(1)Obtain experimental datasets.Collects the "false information" published in the Weibo Community Management Center as the disinformation dataset.The collected content includes the publisher’s information and the basic attributes of comment users and Weibo attributes.At the same time,the data on the mobile terminal of Weibo is collected as non-disinformation.The information dataset together constitute part of the experimental dataset in this study.(2)Uses the bi-directional long short-term memory model with attention mechanism to extract the sentiment feature of the publisher.By analyzing the disinformation text on Weibo social media,it can be found that when a publisher posts a disinformation,it usually contains complex emotions such as anger,sadness,and happiness.This research uses an attention mechanism with a bi-directional long short-term memory model to extraction the publisher’s sentiment feature by labeling some text emotions.(3)Uses the hierarchical attention network model to extract the sentiment feature of comment users.The publisher publishes a false message usually to resonate with netizens.After the publisher publishes a disinformation with complex emotions,users will express their attitudes or opinions on the content of the Weibo text in the comments.Given that a review text under a Weibo is composed of multiple review sentences,and the sentences are composed of words,for this type of review text with a hierarchical structure,this research uses a hierarchical attention network model to identify users’ opinions on the Weibo and attitude.(4)Researches on disinformation recognition considering sentiment feature.After obtaining the sentiment feature of publishers and comment users,combined with traditional disinformation recognition feature,this research proposes a disinformation recognition model considering sentiment feature.The model takes the sentiment feature of the publisher and the comment user as input features,and combines the semantic information of the microblog text feature,the credibility feature of the publisher,and the influence feature of the microblog to realize the identification of disinformation.This paper selects Sina Weibo text data as the experimental dataset.The final experimental results show that the sentiment feature of publishers and comment users considered by this research play an important role in the identification of disinformation,and a disinformation recognition model that considers sentiment features,it can effectively improve the accuracy of disinformation recognition,and has certain academic value and practical application value. |
