Collaborative QoS Prediction via Context-Aware Factorization Machine

With the prevalence of Web/Cloud/IoT services on the Internet, to select service with high quality is of paramount importance for building reliable distributed applications. However, the accurate values of the quality of services (QoS) are usually uneasy to obtain for they are typically personalized and highly depend on the contexts of users and services such as locations, bandwidths and other network conditions. Therefore, personalized and context-aware QoS prediction methods are desirable. By exploiting the QoS records generated by a set of users on a set of services, this paper proposes a collaborative QoS prediction method based on Context-Aware Factorization Machines named CAFM, which integrates the context information of services and users with classic factorization machines. Comprehensive experiments conducted on a real-world dataset show that the proposed method significantly outperforms existing methods in prediction accuracy.