A collaborative framework of web service recommendation with clustering-extended matrix factorisation

QoS-based web service recommendation is an important technique to select suitable services to users. In this paper, we aim to achieve superior recommendation accuracy by leveraging the known QoS records. To achieve this goal, we employ the clustering algorithm and Matrix Factorisation model (MF), and propose a collaborative framework of web service recommendation. Using the clustering algorithm, we cluster users and services into different clusters based on their QoS records, and identify similar cluster centres for each user and each service. We propose two clustering-extended MF models, i.e., service clustering-extended MF model (SC-EMF) and user clusteringextended MF model (UC-EMF). In both models, the QoS values are predicted by two parts. One is the invocation experience of the target service or user, and the other is that of the similar centres. The experimental results show the effectiveness of our models.