Modeling and evaluation of trust with an extension in semantic web

The aim of this paper is two folds. Firstly, some of the well known methods of trust modeling and trust evaluation that relates mainly to the semantic web structure are reviewed and analyzed. A categorization for calculation of trust and an analytical view of possible models of trust rating through a chain of acquaintances are presented. Based on experimental results the well known methods are compared and contrasted. Secondly a new method for evaluating trust is also proposed. This new model has the advantages of simplicity in calculation and enhanced accuracy. The method is associated with two algorithms, an algorithm for propagation and another for aggregation. The propagation algorithm utilizes statistical techniques and the aggregation algorithm is based on a weighting mechanism. The technique is named Max-weight method and is also implemented and the results are compared based on a designed accuracy metric. The proposed method may be employed as a subsystem for trust management in semantic web and trust evaluation in human interaction in a social networks as well as machines (artificial agents). Experimental results illustrate the efficiency and effectiveness of the proposed method.