A task-based methodology for specifying expert systems /
Abstract: "There has been an increasing interest in improving the quality and reliability of AI system [sic]. As a result, several methodologies for specifying expert systems have been proposed in recent years. However, these methodologies have been demonstrated either to be a conceptual lev...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Book |
| Language: | English |
| Published: |
College Station, Tex. :
Texas A & M University, Computer Science Dept.,
[1991]
|
| Series: | Technical report (Texas A & M University. Computer Science Department) ;
91-033. |
| Subjects: |
| Summary: | Abstract: "There has been an increasing interest in improving the quality and reliability of AI system [sic]. As a result, several methodologies for specifying expert systems have been proposed in recent years. However, these methodologies have been demonstrated either to be a conceptual level approach without any proof of concept, or for a particular problem solving architecture or for a specific type of expert systems. It is unclear how one can apply these methodologies to other types of expert systems or other problem solving architectures. To alleviate this problem, we propose a task-based methodology for specifying both the model and the process knowledge of an expert system at different abstraction levels. Specification [sic] are acquired and organized around the general notion of task. The model specification of a task describes a partial model of the application domain and a partial model of the problem solving states relevant to the task. The process specification of a task describes states before, during and after the task, the anticipated behavior of tasks, as well as methods for accomplishing the task. To capture specification at its appropriate abstraction levels, we use the task/method/subtask relationship to organize a specification into a hierarchical multi-layered structure. A piece of high-level abstract specification can thus be refined to a more detailed specification at a lower level. Specification at different abstraction levels can thus be verified for their consistency, completeness, and redundancy. Beginning with the documentation and source code of the existing implementation of R1/SOAR, we constructed a specification (in a reverse engineering fashion), using our task-based methodology. The proposed specification methodology will not only enable the verification of an expert system's knowledge in an earlier phase of the software life cycle, but also facilitate the validation and the maintenance of large expert systems." |
|---|---|
| Item Description: | "July 1991." |
| Physical Description: | 42 leaves : illustrations ; 28 cm. |
| Bibliography: | Includes bibliographical references. |