Big Data Trust Evaluation Based on D-S Evidence Theory and PageRank Model

Abstract

In view of the multi-dimensional attributes and uncertainties existing in the trust evaluation of big data nodes under the big data environment, this paper proposes a big data distributed collaborative trust management framework and a trust assessment model and a trust assessment model. On the basis of big data production environment, the new big data trust processing agent is creatively proposed to calculate and manage data source trust information and establish trust network through trust processing agent. In the calculation of direct trust, this paper adopts the multi-dimensional trust evaluation method of D-S evidence theory to evaluate the direct trust value of the trust agent to the big data source and uses dynamic weight to modify Dempster’s rule of combination to avoid evidence conflict. When calculating the indirect trust value, this paper uses PageRank algorithm to calculate the recommended value of the trust processing agent. Finally, the integrated trust values of big data sources are obtained by combining direct and indirect trust. Experimental analysis shows that the trust model can make trust evaluation for big data sources accurately, has obvious ability to distinguish trust, and has strong robustness.