WPI Computer Science Department
Spring Semester 2010
*** Under Construction ***
CS 538 - Knowledge Based Systems
The course will review major expert knowledge-based problem-solving systems, such as MYCIN, DENDRAL, XCON, AM, etc. It will concentrate on an analysis of the knowledge structures and problem-solving strategies of each system in order to classify and compare these systems. For each system an attempt will be made to evaluate its contribution to our understanding of problems that AI can tackle.
The goals of the course are to expose participants to some of the more influential Expert Systems, and to encourage critical evaluation of both the methods used and the practical and theoretical contributions made by each system. We'll also be looking at the developments in the study of Problem Solving Methods (PSMs).
It is expected that this course will be very useful for those doing or looking for thesis/dissertation topics in Artificial Intelligence, or related fields such as Intelligent Tutoring Systems, Design and Diagnosis.
Participation is limited to a maximum of 15 people. A graduate AI course (CS 534), or its equivalent, or the instructor's permission, is a prerequisite!
As there is no text that exactly matches this course, the course will be based on readings from the literature. However, the selected textbook will be used as background reading. If the class is smaller we will focus on it more.
An initial set of readings will consist of articles that introduce "classic" systems as well as some research into Problem Solving Methods.
The first major portion of the course will be run in participatory seminar style with a different "classic" systems being examined every week. Presentations about each system should be designed to last about an hour. Each person presenting will be provided two weeks in advance with at least the references to the major books/papers describing the system, and, in most cases, they'll be given a folder containing copies of the papers. The rest of the class will receive a paper about the system one week in advance so that they will be prepared for the more detailed presentation. This format is conditional on there being a large enough registration for the class.
Every person making a presentation will be expected to prepare and distribute at the start of the following class a "Profile" of the system(s) they discussed. The format of that is described elsewhere. This is a vital component of this course.
The last portion of the class will consist of readings selected from the recent research literature on Problem Solving Methods, and possibly KBS systems of special interest to the class members.
For most papers read in the class a short critique should be written that notes the key contributions or weaknesses of the paper. A critique should be no more than a page long, and may be in bulleted form.
The final ingredient of this course is a small project, to be handed in prior to the end of the semester. The project is described elsewhere.
Evaluation will be based on each individual's presentation, on the Profile for their system, on class participation, the Project, and the critiques.