Article Preview
TopThere have been a number of recent attempts for the automation of workflow composition process with logical workflow composition, semantic workflow composition, abstract process model, and artificial intelligence (AI) approaches. Some of these works have been attempted for prototype implementation, and yet at the present time, there appears to be very little evidence of testing and validation particularly through field deployment.
As described by Hafner (2009), the manual service workflow architecture has five layers, which include applications layer, web services workflow composition or publication and discovery layer, service description layer, XML messaging layer, and transport layer. Today’s frameworks automate portions of this layered-architecture to ease the pain for developers. For example, the .NET framework automates the service description layer, XML messaging layer, and the transport layer. However, there are still tedious manual tasks that the service developer is expected to complete, including the composition of the services into workflows, and their integration with applications. The following subsections present a literature survey on the attempts for automation of the workflow composition problem. The focus is given to the integration of AI planning techniques with service workflow composition since it appears to have attracted considerable attention from the researchers.