A Spanning Tree Based Approach to Identifying Web Services

• Published in: International journal of web services research 2004, Vol.1 (1), p.1-20
• Main Authors: Jain, Hemant, Zhao, Huimin, Chinta, Nageswara R
• Format: Article
• Language: English
• Subjects: Binary trees (Computers); Genetic algorithms; Graph theory; Identification and classification; Industry forecasts; Internet service providers; Methods; Software; Technology application; Web services; Web sites; World Wide Web
• ISSN: 1545-7362; 1546-5004
• DOI: 10.4018/jwsr.2004010101

Web service has been envisioned as an important trend in application development and integration. It allows pre-built applications/application components wrapped as web services to interact with each other through standardized interfaces and form larger application systems. This paper describes a formal approach to web services identification, which is a critical step in designing and developing effective web services. The approach takes an analysis level object model, representing a business domain, as input and generates potential web service designs, in which the classes in the object model are grouped into appropriate web services based on static and dynamic relationships between classes. An initial hierarchical grouping of classes is derived using a maximum spanning tree algorithm. A set of managerial goals for evaluating alternative designs is derived based on business strategy of web service fabricator. Since the managerial goals are conflicting, a multi-objective genetic algorithm has been designed to search for alternative non-dominated solutions, from which a preferred solution can be selected. The approach has been implemented in a web services identification tool and used for designing web services in an auto insurance claims domain. The experts evaluated the utility of the approach.