An approach to expert control of interactive software systems.

Area of Science:

• Artificial Intelligence
• Expert Systems
• Computational Science

Background:

• Expert problem-solving often combines mathematical techniques with domain-specific experiential knowledge.
• Large software packages commonly house these mathematical techniques for expert use.
• Integrating existing software and expert knowledge into practical expert systems presents a significant challenge.

Purpose of the Study:

• To describe a framework for a hybrid expert system capable of representing problem-solving knowledge.
• To address the challenge of integrating large software packages and expert usage knowledge into expert systems.
• To develop a system that dynamically selects and interprets appropriate programs and parameters for problem-solving.

Main Methods:

• Developed a hybrid expert system architecture combining a production system with mathematical methods.
• Reorganized existing software packages into a mathematical-method representation compatible with the hybrid system.
• Utilized expert knowledge for dynamic selection and interpretation of computational tools and parameters.

Main Results:

• Successfully created a framework for a hybrid expert system.
• Demonstrated the framework's application through the development of the expert log analysis system (ELAS).
• The system effectively uses expert knowledge to guide the application of mathematical methods and software.

Conclusions:

• The proposed hybrid expert system framework offers a viable approach for integrating complex software and expert knowledge.
• This methodology facilitates the creation of practical expert systems for domains requiring sophisticated mathematical analysis.
• The ELAS system serves as a successful proof-of-concept for this hybrid approach in well-log analysis.