Web Service Classification Based on Reinforcement Learning and Structured Representation Learning

In recent years, the rapid growth of the number of web services on the Internet has made it difficult to discover services from a large number of web services and classify them efficiently, which poses a huge challenge for users to provide appropriate web services. The existing web service classification technology has problems such as sparse description text, insufficient consideration of attribute information and structural relationships, so the accuracy of web service classification can be further improved. In response to the above problems, this paper first proposes an information distillation model based on reinforcement learning. Reinforcement learning is introduced to delete redundant information and retain valid information. Then, the classification network classifies the distilled information and determines the reward based on the prediction result. Then, the reward is fed to reinforcement learning for gradient update and the classification network is updated inversely according to the classification result. It can effectively alleviate the inherent noise caused by the introduction of the attention mechanism. It can also learn classification tasks by identifying important words or task-related structures without displaying structural annotations (ie, artificially building models). Through comparative experiments on the real data set of Programmable Web, the results show that the proposed model achieves better results on the benchmark data set than other methods.