Time-Aware and Sparsity-Tolerant QoS Prediction Based on Collaborative Filtering
2016 IEEE International Conference on Web Services (ICWS)(2016)
摘要
Quality of Services (QoS) is an important criterion to evaluate Web services recommendation system. Due to factors including various network conditions, QoS values are dynamic and time-varying. In reality, the data is too spare to fit in with traditional time series forecasting model (e.g., ARIMA). To address this crucial challenge, this paper proposes a novel time-aware and sparsity-tolerant QoS values prediction approach based on collaborative filtering. Our approach combines limited historical QoS value with collaborative filtering method to forecast the personalized QoS values. Based on the limited data, our approach firstly forecasts user-service pairs that have historical usage experiences, and then uses CF-based method to predict personalized QoS values. Finally, we combine the results from temporal forecasting with those from CF prediction as the final forecasted QoS values. The extensive experiments show that the proposed approach efficiently improves the forecasting coverage and accuracy.
更多查看译文
关键词
QoS prediction,time-aware,K-means clustering,collaborative filtering,Web services
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络