Residual attention graph convolutional network for web services classification

More and more attention has been paid to web service classification as it can improve the quality of service discovery and management in the service repository, and can be widely used to locate developers’ desired services. Although traditional classification method based on supervised learning model to this task shows promising results, it still suffered from the following shortcomings: (i) the performance of conventional machine learning methods highly depends on the quality of manual feature engineering; (ii) some classification methods (such as CNN, RNN, etc.) are usually limited to very shallow models due to the vanishing gradient problem and cannot extract more features, which have great impact on the accuracy of web service classification. To overcome these challenges, a novel web service classification model named Residual Attention Graph Convolutional Network (RAGCN) is proposed. Firstly, adding an attention mechanism to the graph convolutional network can assign different weights to the neighborhood nodes without complicated matrix operations or relying on understanding the entire graph structure. Secondly, using residual learning to deepen the depth of the model can extract more features. The comprehensive experimental results on real dataset show that the proposed model outperforms the state-of-the-art approaches and proves its potentially good interpretability for graphical analysis.