In recent years, the credibility of information on social networks has attracted considerable of interest due to its critical role in the spread of information online. In this paper, we argue that the quality of information created on social networks can be analyzed using its provenance data. In particular, we propose an algorithm that assesses information credibility on social networks in order to detect fake or malicious information using a fuzzy analytic hierarchy process to assign proper weights to the proposed metrics. In order to test the usability of the proposed algorithm, we introduce a prototype implementation and test it on a large-scale synthetic social provenance dataset. The initial results reveal a proportional relationship between our proposed distance from positivity algorithm and the provenance graph metrics-based user credibility.