Trustworthiness of Information Sources and Information Pedigrees

To survive, and indeed thrive, in an open heterogenous information sharing environment, an agent's ability to evaluate the trustworthiness of other agents becomes crucial. In this paper, we investigate a procedure for evaluating an agent's trustworthiness as an information source. By separating the procedure into competency analysis and sincerity analysis, we are able to deal with complicated cases involving the passing-on of information, where the same information may reach a receiver agent via different routes. In order to keep information about the source agent we use an information pedigree as a means to maintain the history of communicated information. Our evaluation of trustworthiness can be employed to drive data fusion, weighted knowledge base merging, and multiple source conflict resolution.