Description

Abstract

Service-oriented computing (SOC) paradigm and its realization through standardized web service technologies provide a promising solution for the seamless integration of single-function applications to create new large-grained and value-added services. With the number increasing of Web services, Quality-of- Service (QoS) is usually employed for describing nonfunctional characteristics of Web services. Among different QoS properties of Web services, some properties are user independent and have identical values for different users (e.g., price, popularity, availability, etc.). The values of the user independent QoS properties are usually offered by service providers or by third-party registries (e.g., UDDI). On the other hand, some QoS properties are user dependent and have different values for different users (e.g., response time, invocation failure rate, etc.). Obtaining values of the user dependent QoS properties is a challenging task, since realworld Web service evaluation in the client side is usually required for measuring performance of the user dependent QoS properties of Web services. This paper presents a neighborhood based collaborative filtering approach to predict such unknown values for QoS-based selection.  Proposed methods have introduced three new features to remove the impact of different QoS scale we introduce the adjusted cosine-based similarity calculation method we presents a data smoothing process for improving the prediction accuracy. A similarity fusion approach is used to handle the data sparsity problem. Finally, a two-phase neighbor selection strategy is proposed to improve its scalability.