TRUTHSEEKER: BUILDING THE LARGEST ANNOTATED DATASET FOR CONTENT VERACITY ON SOCIAL MEDIA

A big worldwide problem that has emerged in recent years is the ease and rapidity with which misinformation can travel across social media platforms, with negative consequences for public trust, political discourse, and overall society health. Accurate content authenticity identification algorithms necessitate large, trustworthy datasets to circumvent this issue. The Truth Seeker project is discussed in this research. It aims to compile the largest dataset with annotations regarding the veracity of material across all the main social media platforms. Verifying the legitimacy of numerous social media posts required a multi-pronged approach that included AI-assisted tagging approaches, crowd sourcing verification, and expert fact-checking. For complete analytical modeling, it is necessary to record important metadata. Factors like as user engagement, contextual indicators, timeliness, and reliability of sources are part of this. In this study, the methodology, annotation structure, quality control procedures, and ethical considerations that were part of the development of Truth Seeker are detailed. Results from first experiments employing sophisticated machine learning models demonstrate that the dataset can contribute to the development of better content filtering, false information monitoring, and false information detection systems. Our goal in making Truth Seeker available to academics is to spark fresh thinking on AI reliability, social media regulation, and digital literacy initiatives.