Department of Computer Science
CS 681 Designing Expert Systems
Meeting
time: Thursday 4:30 pm – 7:10 pm
Meeting location: The Engineering Building 5358
Instructor: Dr. Gheorghe Tecuci,
Professor of Computer Science (http://lac.gmu.edu/members/tecuci.htm)
Office
hours: Thursday 3:30 pm – 4:20 pm
Office: The Engineering Building 4613
Phone: 703 993 1722
E-mail: tecuci
at gmu dot edu
Course Description
Prerequisite: CS 580 or permission of instructor
This course presents the theory and
practice of designing and developing systems that rely on expert knowledge and
reasoning to solve complex problems in a specific (scientific, engineering,
medical, military, etc.) domain. Such an expert system or knowledge-based agent
may assist a human expert in complex problem solving and decision-making, may
be used by a non-expert user, or may teach problem solving and decision-making
to a student. Capturing, using, preserving, transferring, and sharing knowledge
is of critical importance to any organization as society evolves from an
information society to a knowledge society. Therefore, the ability to design
and develop such expert agents for a wide variety of domains is a highly
valuable expertise. The course covers both the basics of expert agents and knowledge
engineering as well as advanced research topics, and involves the students in
expert agents research. Basic topics include modeling
expert reasoning, ontology design and development, logic and probabilistic
reasoning, knowledge acquisition and learning, knowledge base verification,
validation and integration. Advanced topics include mixed-initiative reasoning,
agent teaching and multistrategy learning, and collaborative
problem solving.
The students will learn about all
the phases of building an expert agent and will experience them first-hand by
using the Disciple development environment. Disciple has been developed in the Learning Agents Center (http://lac.gmu.edu)
of George Mason University and has been successfully used to build expert agents
for a variety of domains, including: intelligence analysis, military center of
gravity determination, medical diagnosis, website
evaluation, course of action critiquing, emergency response planning, teaching
critical thinking, and PhD advisor selection.
The classes will consist of three
parts: theory, tools and project. In the theoretical part, the instructor will
present and discuss the various phases and methods of building an expert agent.
In the second part the students will experience the use of advanced artificial
intelligence tools for building expert agents. In the project part the students
will design and develop an expert agent in a domain of their choice.
Grading Policy
Exam – 50%
Project (Expert System Development) – 50%
Readings
Tecuci G., Lecture Notes on Designing Expert Systems, Fall
2009 (required).
Tecuci G., Building Intelligent Agents: An Apprenticeship Multistrategy
Learning Theory, Methodology, Tool and Case Studies, Academic Press, 1998
(recommended).
Additional papers required or
recommended by the instructor.
Lecture Notes on Designing Expert
Systems
Introduction
Classical Approaches to the Design and Development of Expert Systems
Ontology Design and Development
Learning-Oriented Knowledge Representation
Problem Reduction and Solution Synthesis
Modeling Expert’s Reasoning
Agent Teaching and Multistrategy Rule Learning
Mixed-Initiative Problem Solving and Knowledge Base Refinement
Tutoring Expert Problem Solving Knowledge
Design Principles for Expert Systems
Frontier Research Problems
Honor
Code
You
are expected to abide by the GMU honor code. Homework assignments and exams are
individual efforts. Information on the university honor code can be found at http://academicintegrity.gmu.edu/honorcode/.
Additional
departmental CS information: http://cs.gmu.edu/wiki/pmwiki.php/HonorCode/CSHonorCodePolicies