QoS-aware temporal prediction model for personalized service recommendation

An optimized temporal prediction model for quality of service (QoS) is proposed to improve the prediction accuracy of personalized service recommendation. A baseline model is proposed to transform the prediction task from overall value prediction to bias value prediction, and combined with the matrix factorization technique to build the baseline matrix factorization (BMF) model. Matrix factorization models are designed to denote the time effect of both client and server sides, and then integrated with the BMF model to build the temporal baseline matrix factorization (TBMF) model. Experimental results show that, compared with the existing temporal prediction model for QoS, the BMF model can improve the precision substantially, and that the TBMF model can be improved further.