A Collaborative Filtering Method for Cloud Service Recommendation via Exploring Usage History

Abstract

With the rapid development of cloud computing technique, cloud service recommendation has attracted significant research interest nowadays. The main task of cloud service recommendation is to provide a list of new cloud services for users to satisfy their requirements and the key of accurate recommendation is the complete and accurate identification of users’ preference. In this study, we propose a cloud service recommendation approach based on collaborative filtering via exploring user usage history. This approach first computes user similarity using an improved cosine similarity method, which is adjusted by cloud service popularity. Then several similar users are selected as user neighbors. At last, it predicts the possibility that active users invoke candidate services according to user neighbors. Experimental results show that the proposed approach performs better than existing approaches in terms of recommendation accuracy.