Truth Discovery in Big Data Social Media Sensing Applications

Abstract

The detection of truthful information amid data provided by online social media platforms (e.g., Twitter, Facebook, Instagram) is a critical task in the trend of big data. Truth Discovery is nothing but the extraction of true information or facts from unwanted and raw data, which has become a difficult task nowadays in today's day and age due to the rampant spread of rumors and false information. Before posting anything on the social media platform, people do not consider fact-checking and the source authenticity and frantically spread them by re-posting them which has made the detection of truthful claims more difficult than ever. So, this problem needs to be addressed soon since the impact of false information and misunderstanding can be very powerful and misleading. This mission, truth discovery, is targeted at establishing the authenticity of the sources and therefore the truthfulness of the statements that they create without knowing whether it is true or not. We propose a Big Data Truth Discovery Scheme (BDTD) to overcome the major problems. We have three major problems, the main one being "False information spread" where a large number of sources lead to false or fake statements, making it difficult to distinguish true statements, now this problem is solved by our scheme by studying the various behaviors of sources. On Twitter for example rumormongering is common. The second problem is "lack of claims" where most outlets contribute only a tiny small number of claims, giving very few pieces of evidence and making it not sufficient to analyze the trustworthiness of such sources, this problem is addressed by our scheme where it uses an algorithm that evaluates the claim’s truthfulness and historic contributions of the source regarding the claim. Thirdly the scalability challenge, due to the clustered design of their existing truth discovery algorithms, many existing approaches don't apply to Big-scale social media sensing cases so this challenge is managed by our scheme by making use of frameworks HTCondor and Work Queue. This scheme computes both the reliability of the sources and, ultimately, the legitimacy of statements using a novel approach. A distributed structure is also developed for the implementation of the proposed scheme by making use of the Work Queue (platform) in the HTCondor method (maybe distributed). Findings of the test on a real-world dataset indicate that the BDTD system greatly outperforms the existing methods of Discovery of Truth both in terms of performance and efficiency.