QoS prediction for service selection and recommendation with a deep latent features autoencoder

The number of services on the Internet has increased rapidly in recent years. This makes it increasingly difficult for users to find right from a large functionally equivalent candidate. In many cases, invoked by user is quite limited, resulting missing QoS values and sparseness data. Consequently, predicting important exact service among similar services. However, improving accuracy prediction still problem. Despite successful results proposed methods, there are set issues that should be addressed, such as Sparsity Overfitting. To address these improve accuracy. this paper, we propose novel framework reduce error. named auto-encoder neighbor features (Auto-NF) consists three steps. first step, an extended similarity computation method based Euclidean distance compute between neighbors. second form clusters neighbors partition initial matrix into sub-matrices data sparsity third simple neural network autoencoder can learn deep select ideal latent factors overfitting phenomenon. validate evaluate our method, conduct series experiments use real dataset with different densities. experimental demonstrate achieves higher compared existing methods.