AFlow: An Automated Web Services composition system based on the AI planning and workflow

Automated Web Service composition is an important way for integrating information flexibly and quickly. However, existing composition approaches are insufficient for satisfying users' requirements. In this paper, we firstly discuss the limitations of the existing web service composition approaches and then introduce the separation model to separate the representation of web service type definitions from instance definitions in order to organize the expected thousands of web services well. At last we construct an automated web service composition system, which allows for fast and flexible composition of web services. This system AFlow consists of two subsystems: AI Planning subsystem and Service Matchmaking subsystem. The AI Planning subsystem receives the user's request to form an abstract workflow and transfer it to the Service Matchmaking subsystem. The Service Matchmaking subsystem selects the most suitable service instances according to the abstract workflow via the matchmaking engine and then generates an executable workflow to execute.