| Abstract: | One of the major problems faced by developers of expert systems is the identification of reusable problem solving designs. Often in the development of a new expert system project a required problem solving design used previously can be applied to the new project. However, time is expended in recreating the same design because of a lack of reusable tools. This research provides a method for retrieving reusable problem solving knowledge by modeling expert systems using a principled modeling language. Principled knowledge representation schemes have been used to model complex software systems. However, the potential for applying these principled modeling techniques for explicitly capturing the functional requirements of expert systems has not been fully explored. This creates difficulties in capturing an abstract view of the requirements and the functionality of an expert system's modules. This thesis uses an Artificial Intelligence-based knowledge representation scheme for improving the reusability of problem solving designs. This research improves the reusability of problem solving designs by: (1) developing a principled modeling method that has a high-level view of the system requirements and functionality of an expert system. (2) facilitating the reuse of problem solving knowledge, especially the classification of software components. (3) formulating a principled-similarity measure to analyze the similarity among the functional specifications in a software repository. (4) providing an intelligent retrieval algorithm for extracting reusable candidates from the repository. To achieve these objectives, this thesis uses an approach that (1) specifies the requirements and functionality of an expert system. (2) defines meaningful descriptions of software components to aid the classification and analysis process. (3) uses a principled and consistent similarity measure. (4) retrieves reusable software component. In brief, this research explores the feasibility and benefits of using principled modeling techniques to capture explicitly the functionality of components of expert systems to improve the reusability of problem solving knowledge. The contributions of this research are the enhancement of expert system component reusability through semantic-based similarity analysis and retrieval of expert system components. |
|---|