ST-LDA: High Quality Similar Words Augmented LDA for Service Clustering

Abstract

Service discovery is a key problem in the field of services computing, which is essential to improve the accuracy and efficiency of both services composition and recommendation. Service clustering is a major way to facilitate service discovery. The main technical difficulty in solving service clustering problem lies in the semantic gap among services. Some traditional approaches like LDA perform well in service clustering to some extent. However, their performances are still limited by the inevitable semantic noise words. To bridge this gap, we propose a novel solution, namely ST-LDA (short for “Similar Words and TF-IDF Augmented Latent Dirichlet Allocation”), approaching the challenges from the perspective of similar words learning and noise words filtering to improve service clustering. Specifically, we adopt Word2Vec to adapt the representation of services, and learn a list of similar words in service corpus. Moreover, we further integrate TF-IDF into our similarity calculation to filter noise words. In this way, we can enhance LDA with the similar words finding and filtering strategy for service clustering. We conduct extensive experiments on a real-world dataset, which demonstrate that our approach can improve the efficiency of service clustering.