From Path to Individual: A Distributed Reputation Detection Model for Information Diffusion

In previous computational reputation models, the reputations rely heavily on the direct interaction experiences. However, in some cases lacking of direct interactions (e.g. information diffusion), the interaction reputations can be hardly obtained in a distributed manner. To solve such problem, a novel distributed reputation detection model following the pattern "From Path to Individual" is proposed in this paper. The proposed model can produce individual reputations by evidences of path reputations which are actually the perception of concatenation of agents\´ actions. The obtained individual reputations can be considered as an alternative for the interaction reputations in the cases lacking of direct interactions due to their same information source (perception of agents\´ actions). This model consists of two components, one for the transformation from obtained path reputations to individual reputations of agents and the other one for the detection of path reputations in information diffusion scenario. Through experiments, our model can be validated to be efficient and reliable for agent\´s reputation detection in terms of detection coverage and accuracy.