Reasoning about user trustworthiness with non-binary advice from peers

In this paper we outline an approach for reasoning about the trustworthiness of users when advice from peers is provided and a majority opinion is central to the overall calculation. Typically, trust modeling frameworks have reasoned with binary reports from peers (i.e. trustworthy or not). In this paper, we illustrate how to do trust modeling when numeric advice is provided, instead (i.e. degree of trustworthiness within a continuum). This is done in the context of a specific application: directing the travel decisions of users with peer-provided traffic reports. We demonstrate the effectiveness of our solution by mapping the average time for completing paths in simulated traffic environments when vehicles are modeling trust using our framework and road reports are shared. We conclude with a discussion of the value of our approach for other applications as well.