In partial fulfillment of the Requirements for the Degree of
Master of Science

will defend her thesis

The goal of the project A Learning Program System (ALPS) is to develop a system that mimics a human being in its learning. This involves building a knowledge learning system that can learn all kinds of knowledge and solve problems based on learned knowledge. It is based on a simple building block approach that more complicated knowledge is based on simpler knowledge.

The hierarchy learning subsystem described in this thesis is a part of the ALPS learning system. It is motivated by the fact that most kinds of knowledge are involved in hierarchies, so the work should only be done once, and used when needed. This thesis identifies the common functionality of maintaining hierarchies, and develops a new class with the functionality so that it can be used in learning all kinds of knowledge that have hierarchies. It demonstrates the usefulness of this newly developed class in learning the geographical containment hierarchy for objects. The geographic containment hierarchy contains two inter-related hierarchies, one in the abstraction level, and the other in the object level. In addition, this new class is also used in implementing the generalization/specialization hierarchies in many kinds of knowledge, such as data types, polygons, and abstractions.

The design of the system has been established using object-oriented paradigms and the solution is expressed in Unified Modeling Language (UML) notations. The system is implemented in C++. The subsystem performs consistency checks while learning the hierarchies, and answering questions posted by the user.

Date: Wednesday, April 21, 2004

Time: 02:30 P.M

Place: 550-PGH

Faculty, students, and the general public are invited.

Thesis Advisor: Dr. Kam-Hoi Cheng