A knowledge coverage-based trust propagation for recommendation mechanism in social network group decision making

Abstract

Trust is a typical relationship in social network, which in group decision making problems relates to the inner relationship among experts. To obtain a complete trust relationship of a networked group of experts, firstly, a novel knowledge coverage-based trust propagation operator is proposed to estimate the trust relationship between pairs of unknown experts. The novelty of this trust propagation operator resides in its account of the domain knowledge coverage of experts. Desirable properties regarding boundary conditions, generalisation and knowledge coverage absorption are studied. The comparison with existing operators of boundary conditions shows the rationality of the proposed operator. Next, a knowledge coverage-based multi-paths trust propagation model for constructing complete trust network is investigated. The proposed approach aggregates all trust paths to collect all trust information and penalise trust decay. Secondly, a trust order induced recommendation mechanism is proposed by combining subjective and objective weights. Thus, experts can accept consensus recommendations by subjective and objective trust. This recommendation mechanism allows the inconsistent experts to accept the advices they trust. The validity and rationality of the proposed recommendation mechanism is mathematically proved, and a numerical example is utilised to illustrate the calculation process of the proposed method.