Toward End-User Specification and Design of Business Systems
This dissertation research is a collection of three essays that follow a design science approach to develop a contemporary, web service repository that can be used to aid the elicitation of business software requirements. The first essay presents an evaluation framework called the coalescence model, which defines specific design characteristics to guide the evaluation of software component repositories. The model provides an evaluation rubric as well as objective measures of coalescence and balance that can be used for comparing the relative efficiency and effectiveness among repositories. The model indicates that existing repositories exhibit limitations when they are used to aid in the requirements elicitation process. The second essay discusses the design characteristics presented in the first essay in light of creating a repository to facilitate both the design and delivery of contemporary web services. The facet schema of the repository is extended to include a workflow and composition facet, which graphically represent two unique qualities of contemporary web services. A prototype of the repository is described, created, and evaluated. The prototype is shown to be efficient and effective at both design and delivery of contemporary web services, and the prototype has higher coalescence and balance scores than any of the existing repositories. The third essay extends the design of the prototype repository by enabling the repository to function as a guide and stimulus to enable business users to discover and capture requirements in an iterative fashion. The repository is given the capability to match a captured list of requirements with an appropriate set of web services that could be later assembled into a composite application. The matching algorithm employs a hierarchical approach to locate entire service compositions at the highest granularity level possible, while minimizing cost and complexity of later deployment. The matching algorithm also utilizes a semantic matching approach based on the Explicit Semantic Analysis (ESA) algorithm to calculate term similarity when matching individual requirements with web services (Gabrilovich and Markovitch 2007). The extension to the Web Service Crawler is then empirically evaluated, and is shown to provide an efficient and effective tool for requirements elicitation.