Service Data Correlation Modeling and Its Application in Data-Driven Service Composition

Zhifeng Gu,Bin Xu,Juanzi Li

DOI: https://doi.org/10.1109/tsc.2010.22

IF: 11.019

2010-01-01

IEEE Transactions on Services Computing

Abstract:In this paper, we propose the Service Data Link model (SDL), a service relationship modeling schema, to describe service data correlations, which are data mappings among the input and output attributes of services. SDL recognizes the close correspondence between service data correlations and webpage hyperlinks, and defines service data correlations with explicit declarations, making it more expressive than the implicit method. We developed an XML implementation for SDL that can be seamlessly integrated into WSDL, the primary web services modeling language nowadays, and serves as an extension of metadata of services interfaces. An application of the SDL model in the domain of data-driven automatic service composition is then presented. First, we combine SDL with the Service Dependency Graph domain model developed by Liang, and present SDG+, our enhanced model which extends the expressive power of SDG to include attribute quantifiers, attribute transforms, and explicit dependencies. Then, we show how SDG+ can be used to improve the performance of composition algorithms in this domain.