Department of Computer Science
IT 803 Instructable Agents
Meeting time: Monday
Meeting location: ST-II, Room 430A
Instructor: Dr. Gheorghe Tecuci, Professor of Computer Science
Office: ST-II, Rm. 421
E-mail: tecuci@ gmu.edu
Prerequisite: CS785 Knowledge Acquisition and Problem Solving
This course is a direct continuation of the course CS785 Knowledge Acquisition and Problem Solving, and relies heavily on the material taught in that course (see the lecture notes of CS785 at http://lalab.gmu.edu/cs785).
Because no lecture can substitute the subtle learning experience gained by effectively building an end-to-end agent, this course will engage the students in developing an agent that assists a person in choosing a PhD thesis advisor. A main objective of this course is therefore to provide hands-on experience with developing instructable agents, based on the principles and methods studied in CS785, and using the Disciple shell as the agent development environment. Each student will develop a personal agent that will assess a potential PhD thesis advisor with respect to a different set of criteria. During this process the students will gain valuable insight into the modeling of the problem solving process, ontology design and development, learning and problem solving. All these agents will be designed to share a common object ontology and will finally be integrated into a more powerful agent.
The class sessions will consist of discussions of the subtleties of various aspects of agent development, guiding the students to develop their personal agents, as part of their weekly assignments. Participation in these discussions is therefore very important and reflected in the final grade.
In addition to developing a personal agent, each student will have a separate project that will consist in writing a publishable research paper focusing on one aspect of the agent development process. This aspect will have to be illustrated with examples from the Ph.D. advisor selection domain.
The course will conclude with discussions of the strengths
and weaknesses of the agent development process, lessons learned, and future
developments. As an end result, it will also produce a detailed documentation of
the agent development process, and a sample agent.
Class participation – 30%
Assignments – 30%
Project – 40%
Lecture notes and various papers indicated by the