SmartVote: a full-fledged graph-based model for multi-valued truth discovery

In the era of Big Data, truth discovery has emerged as a fundamental research topic, which estimates data veracity by determining the reliability of multiple, often conflicting data sources. Although considerable research efforts have been conducted on this topic, most current approaches assume only one true value for each object. In reality, objects with multiple true values widely exist and the existing approaches that cope with multi-valued objects still lack accuracy. In this paper, we propose a full-fledged graph-based model, SmartVote, which models two types of source relations with additional quantification to precisely estimate source reliability for effective multi-valued truth discovery. Two graphs are constructed and further used to derive different aspects of source reliability (i.e., positive precision and negative precision) via random walk computations. Our model incorporates four important implications, including two types of source relations, object popularity, loose mutual exclusion, and long-tail phenomenon on source coverage, to pursue better accuracy in truth discovery. Empirical studies on two large real-world datasets demonstrate the effectiveness of our approach.