PROJECT TITLE :

Posterior-neighborhood-regularized Latent Factor Model for Highly Accurate Web Service QoS Prediction

ABSTRACT:

Because similar users typically have a comparable Quality of Service (QoS) when making use of similar services, neighborhood regularization is of the utmost importance for a latent factor (LF)-based Quality-of-Service (QoS)-predictor. The currently used neighborhood-regularized LF models are dependent on previous information on the neighborhood obtained from either general raw QoS data or geographical information. The former requires additional geographical information, which is typically difficult to collect due to information security, identity privacy, and commercial interests in real-world scenarios. On the other hand, the latter suffers from low prediction accuracy due to the difficulty of constructing the neighborhood based on incomplete QoS data. This study proposes a posterior-neighborhood-regularized latent factor (PLF) model for quality of service (QoS) prediction as a solution to the problems described above. The fundamental concept is to break down the process of LF analysis into three distinct stages: a) primal LF extraction, in which the LFs are extracted to represent involved users/services based on known QoS data; b) posterior-neighborhood construction, in which the neighborhood of each user/service is accomplished based on similarities between their primal LF vectors; and c) posterior-neighborhood-regularized LF analysis, in which the objective function is regularized PLF outperforms other models that are considered to be state-of-the-art in terms of both its accuracy and its efficiency, as shown by the results of experiments conducted with large-scale QoS datasets.