Title:SmartMTD: A Graph-Based Approach for Effective Multi-Truth Discovery

Abstract:The Big Data era features a huge amount of data that are contributed by numerous sources and used by many critical data-driven applications. Due to the varying reliability of sources, it is common to see conflicts among the multi-source data, making it difficult to determine which data sources to trust. Recently, truth discovery has emerged as a means of addressing this challenging issue by determining data veracity jointly with estimating the reliability of data sources. A fundamental issue with current truth discovery methods is that they generally assume only one true value for each object, while in reality, objects may have multiple true values. In this paper, we propose a graph-based approach, called SmartMTD, to unravel the truth discovery problem beyond the single-truth assumption, or the multi-truth discovery problem. SmartMTD models and quantifies two types of source relations to estimate source reliability precisely and to detect malicious agreement among sources for effective multi-truth discovery. In particular, two graphs are constructed based on the modeled source relations. They are further used to derive the two aspects of source reliability (i.e., positive precision and negative precision) via random walk computation. Empirical studies on two large real-world datasets demonstrate the effectiveness of our approach.