Automatic Sampling of Web Services

Abstract

With the widespread of Service-oriented Computing (SOC) and the increasing number of available web services in several domains, service discovery has become one of the main challenges in SOC. Lack of rich service descriptions is one of the several factors that exacerbate this challenge. Therefore, additional information about web services is required. Several approaches and sources have been proposed in the community to gather this information, such as domain experts, service providers, service consumers, etc. However, the increasing number of web services requires automatic approaches and additional sources of information. In this paper, we introduce a novel approach to generate annotations for web services, e.g., tags, by sampling their automatic invocations. The generated annotations are integrated in web forms that are used for future calls of these web services. Providing correct values for input parameters of web services is one of the main challenges involved in this work. We use four sources to assign values for input parameters, namely, random values, outputs of other operations within the same web service, other web services, and external data sources, e.g., DBpedia, Word Net. We have implemented our approach in the context of our service registry and validated it through several experiments.