In Partial Fulfillment of the Requirements for the Degree of
Master of Science
Will defend her thesis
The ability to learn is a distinguishing factor of the human race. Machine learning is an attempt to replicate the learning processes of humans. Previously, A Learning Program System (ALPS) investigated methods of teaching computers various types of knowledge; however, the focus of the system has recently changed from representing knowledge to logical reasoning. Various types of reasoning involve subject matters relating to cause and effects. In order to perform this type of reasoning, one needs to understand the causes and recognize their effects. Knowledge about action is an example of a body of knowledge that involves cause and effect. The focus of this thesis studies learning knowledge about various actions.
We have classified actions into three basic categories: define, simple, and composite. Several actions of each category are learned, and we use the communication capability of the learning program to demonstrate that the knowledge has been learned correctly. Specifically, given a declarative sentence that uses an action verb, our program is able to understand the various effects of the specified action on the appropriate objects involved in the action. The program will also execute those effects to update the knowledge base at the user’s request. The Object-Oriented paradigm is used in the development of this thesis.