Gaussian LDA and Word Embedding for Semantic Sparse Web Service Discovery

In recent years, more and more Web services are published in API marketplaces founded by cloud service providers or third party registries. In this situation, users rely heavily on the search engine model to retrieve their expected Web services. However, due to the fact that Web services registered in API marketplaces are described in short texts, the search engine based discovery method suffers from the semantic sparsity problem, which in turn leads to a poor recall during service discovery. To address this issue, in this paper, we propose a novel Web service discovery approach that uses Gaussian Latent Dirichlet Allocation (Gaussian LDA) and word embedding. More specifically, instead of clustering Web services like most existing service discovery approaches, we use word embedding to map the words as continuous word embeddings to extend and enrich the semantics of service descriptions. We also leverage the Gaussian LDA in service discovery, which takes continuous word distribution as the input and interprets the Web service description as a hierarchical model by its two distributions. Based on the Gaussian LDA and word embedding, we propose a Web service query and ranking approach. Experiments conducted on a real-world Web service dataset demonstrate the effectiveness of the proposed approach.