Trust management framework for social networks

摘要

Inspired by the similarities between human trust and physical measurements, we propose a new system of trust metrics, composed by impression and confidence, which captures both human trust level and its uncertainty, while being intuitive and user friendly. Furthermore, based on measurement error propagation theory, we propose an evaluation framework to compute confidence according to impression transitivity and aggregation operations. Our framework can be used in all applications where human trust is involved, including various types of social networks. Experiments on a real social network validate our framework, as well as the enormous potential of our trust framework in various social network applications. For example, in an experiment on real data we increased 2250 times the “trust view,” while keeping the same level of confidence.