A Granular Approach to Source Trustworthiness for Negative Trust Assessment

The problem of determining what information to trust is crucial in many contexts that admit uncertainty and polarization. In this paper, we propose a method to systematically reason on the trustworthiness of sources. While not aiming at establishing their veracity, the method allows creating a relative reference system to determine the trustworthiness of information sources by reasoning on their knowledgeability, popularity, and reputation. We further propose a formal rule-based set of strategies to establish possibly negative trust on contradictory contents that use such source evaluation. The strategies answer to criteria of higher trustworthiness score, majority or consensus on the set of sources. We evaluate our model through a real-case scenario.