Constructive Representation of Trust: Single Rule Paradigm

A constructive computational framework for trust and reputation assessments is presented. It is proven free from any inconsistent or contradictory assessments under any scenarios of its application. A prototype implementation has been developed. The framework focuses on a single information-theoretical rule as inference mechanism, thus avoiding any biases or spurious constraints in the solutions. The users of our model will find its results intuitively plausible, free from clustering or drift to the extrema. The entire framework is suited for a direct use in economic, financial and intelligence analyses. Keywords-trust; belief revision; maximum entropy; reputation.