Similar Web Services Discovery and Matching Based on P2P and Topic Model Learning

Abstract

We present an enhanced system for similar Web services discovery and matching, which extends our previous work pService. Two steps would be carried out for Web services discovery and matching. Firstly we construct a P2P Web services overlay, which would be applied for extensive search with keywords and got the candidate services sets. This overlay supports similarity search with its locality-preserving feature based on modified Skip Graph and HSFC (Hilbert Space Filling Curve). Secondly we promote a Topic Model based Web services matching algorithm selecting those accurate services from the candidate ones inquired from P2P overlay. We present the experiments on service discovery with overlay and matching accuracy, which shows our approach achieves considerable efficiency.