An Efficient Similarity-Based Model for Web Service Recommendation

Abstract

With the emergence of Cloud and Mobile Computing paradigms which adopt web services (WS) as a means of management, companies worldwide are actively deploying web services within their business environments. Consequently, designing effective Web service recommendation mechanisms is receiving more research attention. Traditional Neighborhood-based Collaborative Filtering (CF) models fail to capture the actual relationships between users and services due to data sparsity. In contrast, Random Walk (RW) algorithm, which is categorized as a sparsitytolerant recommendation approach, suffers from poor recommendation accuracy. In this paper, we first propose an Integrated-Model QoS-based Graph (IMQG), in which users and services represent nodes while weighted Quality of Service (QoS) magnitudes and User/Service similarity measurements serve as edges. Variants of Jaccard coefficient are used to separately compute these similarities. Then, Top-k Random Walk algorithm is applied to generate a final recommendation list. Further, we reduce the model by selecting only a subset of users to better guide RW algorithm. Comprehensive experiments are conducted on a real-world dataset where analysis of results shows high improvement in recommendation accuracy with more tolerance to data sparsity. Utilizing the improved model, a considerable reduction in computation time is achieved while maintaining the recommendation accuracy.