Research on Semantic Web Services Discovery Mechanism Based on Kernel Batch SOM Neural Network

With the rapid growth and wide application of Web services, the research on how to accurately, efficiently and rapidly find the desired Web services has become a challenging subject. In order to improve the efficiency and precision for Web service discovery, a semantic Web services discovery framework based on Kernel Batch SOM neural network is proposed. Firstly, by introducing the WordNet and Latent Semantic Index (LSI) into the VSM lexical vectors to extend semantics and reduce the dimension, the resulting VSM semantic vectors can well describe Web services true semantic characterization; Secondly, by using the kernel trick to modify Regular Batch SOMs weight updating rule, a kernel Batch SOM neural network is proposed to cluster Web services automatically; Thirdly, a kernel Cosine-based similarity matching mechanism is presented to well estimate the similarity of Web services. Finally, the experiments performed on the real-world Web services collection demonstrate the feasibility of the proposed approaches.