On robust truth discovery in sparse social media sensing

Abstract

In the big data era, it's important to identify trustworthy information from an influx of noisy data contributed by unvetted sources from online social media (e.g., Twitter, Instagram, Flickr). This task is referred to as truth discovery which aims at identifying the reliability of the sources and the truthfulness of claims they make without knowing either of them a priori. There are two important challenges that have not been well addressed in current truth discovery solutions. The first one is 'misinformation spread' where a majority of sources are contributing to false claims, making the identification of truthful claims difficult. The second challenge is 'data sparsity' where sources contribute a small number of claims, providing insufficient evidence to accomplish the truth discovery task. In this paper, we developed a Robust Truth Discovery (RTD) scheme to address the above two challenges. In particular, the RTD scheme explicitly quantifies different degrees of attitude that a source may express on a claim and incorporates the historical contributions of a source using a principled approach. The evaluation results on two real world datasetsshow that the RTD scheme significantly outperforms the state-of-the-art truth discovery methods.