A Preference and Honesty Aware Trust Model for Web Services

Trust is one of the most critical factors for a service requestor when selecting the best one from a large pool of services. However, existing web service trust models either do not focus on satisfying customer´s preference for different quality of service (QoS) attributes, or do not pay enough attention to the impact of malicious ratings on trust evaluation. To address these gaps, a dynamic trust evaluation model considering customer´s preference and false ratings is proposed in this paper. The model introduces an approach automatically mining customer´s preference from their requirements. The preference is used to determine the weights on each QoS attribute when integrating trust of the multi-dimensional QoS attributes. The local trust of a service for the customer is derived by combining trust of QoS attributes and customer´s ratings. Then, the customers are divided into different groups according to their preferences, and the honesty of each group is assessed by filtering out dishonest customers based on a hybrid approach combining rating consistency clustering and average method. Finally, the weight on ratings is dynamically adjusted according to the results of honesty assessment when calculating the global trustworthiness of a service for the user group. The simulation results indicate that the model works well on personalized evaluation of trust, and it can effectively dilute the influence of malicious ratings.