A Framework for Efficient Web Services Discovery

Abstract

As number of web services is increasing day by day, it becomes essential to select the correct kind of web service for a user requirement. This is really a challenge to select a web service by understanding its capabilities provided in its description. The selection process of a web service depends upon the mechanism used for discovery and selection of the web service, which is a key component in service-oriented architecture. Some of the existing methodologies for web services discovery cater to web services with semantic annotations i.e. web services that have related semantic descriptions and some are merely based on syntactic based approaches. Selecting an appropriate service out of the discovered services poses a challenge to users. Ranking plays an important role in selecting a desired web service and there exists a number of ways for ranking the web services. In this thesis, we present a framework for finding the most suitable web service according to user s requirements and improve the discovery process. A framework called Automatic Classification and Ranking of Web Services (ACRWebS) has been proposed and designed to deal with the problem of selecting the best web service based on the users need. Our approach uses WordNet based semantic similarity of web services and the user query. Path-length based similarity algorithm has been used in order to find a similarity score between user query and web service description data. It has been observed that the similarity score proposed by the framework meets the expectations of human interpretation. Performance of the ACRWebS has been evaluated and tested using statistical measures. In the proposed framework, classification techniques have been combined with ranking techniques to find the most relevant service for a user (consumer). A dataset of the active web services has been created for conducting experiments. We have also taken another existing test dataset of a researcher as a benchmark for the comparison of experimental results. newline