.png){kind=icon}
.png){kind=icon}
Return
Truth Discovery From Multiple Dependent Sources
.png){kind=icon}
PRE
AI
DOI：10.1109/TCSS.2026.3657429.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Recently, the widespread use of smart devices for Internet of Things has led to a massive growth in data and information on the World Wide Web. However, multiple sources from the Web often provide conflicting descriptions for the same objects, thereby complicating the task of truth discovery, especially when sources may copy information from others. Existing approaches typically neglect the dependence and accuracy of these sources, resulting in low accuracy and efficiency in truth discovery. To solve the problem, we propose <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepenBaye</i>, a source-dependent truth discovery framework, which incorporates Bayesian probability, the Simulated Annealing method, and the expectation maximization (EM) method. By evaluating the copy probability of sources based on false claims, reliable sources are identified with a simulated annealing method to improve efficiency. According to the EM method, source reliability and claim confidence are iteratively calculated to discover the latent truth. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepenBaye</i>’s performance has been validated through extensive experiments, which outperforms existing approaches in efficiency and effectiveness.
Keywords:
Bayesian model
dependent sources
probabilistic graphical models (PGMs)
truth discovery
Journal
IF:
4.9
Papers:
577
Citations:
6.8K
View more organizations