Directory services are a genuine constituent of any distributed architecture which facilitate binding attributes to names and then querying this information, that is, announcing and discovering resources. In such contexts, especially in a business environment, Quality of Service (QoS) and non-functional properties are usually the most important criteria to decide whether a specific resource will be used. To address this problem, we present an approach to the semantic description and discovery of web services which specifically takes into account their QoS properties. Our solution uses a robust trust and reputation model to provide an accurate picture of the actual QoS to user. The search engine is based on an algebraic discovery model and uses adaptive query-processing techniques to parallelise expensive operators. Architecturally, the engine can be run as a centralised service for small-scale environments or can be distributed among any number of cooperating registry providers.

Abstract. Ranking systems such as those in product review sites and recommender systems usually use ratings to rank favorite items based on both their quality and popularity. Since higher ranked items are more likely selected and yield more revenues for their owners, providers of unpopular and low quality items have strong incentives to strategically manipulate their ranking. This paper analyzes the adversary cost for manipulating these rankings in a variety of scenarios. Particularly, we analyze and compare the adversarial cost to attack ranking systems that use various trust measures to detect and eliminate malicious ratings to systems that use no such trust mechanism. We provide theoretical results showing the relation between the capability of the trust mechanism in detecting malicious ratings and the minimum adversarial cost for successfully changing the ranking. Furthermore, we study the impact of sharing trust information between ranking systems to the adversarial cost. It is proved that sharing information between two ranking systems on common user identities and malicious behaviors detected can significantly increase the minimum adversarial cost to successfully attack the two systems under certain assumptions. The numerical evaluation of our results shows that the estimated adversary cost for manipulating the item ranking can be made significant when proper trust mechanisms are employed or combined.

No abstract