Hybrid Fake Information Containing Strategy Exploiting Multi-Dimensions Data in Online Community

Abstract

It is well-established that, in the past few years, internet users have rapidly increased. Meanwhile, various types of fake information (such as fake news or rumors) have been flooding social media platforms or online communities. The effective containing or controlling of fake news or rumor has drawn wide attention from areas such as academia to social media platforms. For that reason, numerous studies have focused on this subject from different perspectives, such as employing complex networks and spreading models. However, in the real online community, misinformation usually spreads quickly to thousands of users within minutes. Conventional studies are too theoretical or complicated to be applied to practical applications, and show a lack of fast responsiveness and poor containing effects. Therefore, in this work, a hybrid strategy exploiting the multi-dimensional data of users and content was proposed for the fast containing of fake information in the online community. The strategy is mainly composed of three steps: the fast detection of fake information by continuously updating the content comparison dataset according to the specific hot topic and the fake contents; creating spreading force models and user divisions via historical data, and limiting the propagation of fake information based on the content and user division. Finally, an experiment was set up online with BBS (Bulletin Board System), and the acquired results were analyzed by comparison with other methods in different metrics. From the extracted results, it has been demonstrated that the proposed solution clearly outperforms traditional methods.