Knowledge-Based Trust: Estimating the Trustworthiness of Web Sources
This addresses the challenge of assessing web source reliability for users and applications, offering a novel approach that could improve information quality online.
The paper tackles the problem of evaluating web source trustworthiness by proposing a method that estimates trust based on the correctness of factual information extracted from sources, rather than traditional hyperlink analysis. It applies this method to 2.8 billion facts from 119 million webpages, with manual evaluation confirming its effectiveness.
The quality of web sources has been traditionally evaluated using exogenous signals such as the hyperlink structure of the graph. We propose a new approach that relies on endogenous signals, namely, the correctness of factual information provided by the source. A source that has few false facts is considered to be trustworthy. The facts are automatically extracted from each source by information extraction methods commonly used to construct knowledge bases. We propose a way to distinguish errors made in the extraction process from factual errors in the web source per se, by using joint inference in a novel multi-layer probabilistic model. We call the trustworthiness score we computed Knowledge-Based Trust (KBT). On synthetic data, we show that our method can reliably compute the true trustworthiness levels of the sources. We then apply it to a database of 2.8B facts extracted from the web, and thereby estimate the trustworthiness of 119M webpages. Manual evaluation of a subset of the results confirms the effectiveness of the method.