Yun-Sun Kang's PhD Thesis Abstract

Knowledge Base Acquisition for a Japanese Language Intelligent Tutoring System

Ph.D., Purdue University, May 1994

Major Professor: Anthony A. Maciejewski

This work describes the development of a student model that is used in a Japanese language intelligent tutoring system to assess a pupil's proficiency at reading technical Japanese. One of the goals of this work is to develop a model representing a student's proficiency in reading one of the distinct orthographies of Japanese, known as katakana. While the effort required to memorize the relatively few katakana symbols and their associated pronunciations is not prohibitive, a major difficulty in reading katakana is associated with the phonetic modifications which occur when English words which are transliterated into katakana are made to conform to the more restrictive rules of Japanese phonology. To accomplish this goal an algorithm is first developed to automatically generate a katakana to English dictionary from raw Japanese text and its English translation. This dictionary is then used as input into a second algorithm that is used to generate the phonological rules that govern the transformations that English words undergo when being transliterated into katakana. This set of phonological transformation rules is treated as the domain knowledge base that the student must acquire in order to become proficient at reading katakana. A student model is designed to assess a student's proficiency, and then appropriately individualize the student's instruction.

The other goal of this work is to develop a model representing a student's proficiency in the grammar associated with reading technical Japanese. To accomplish this goal, first, a computer-assisted knowledge acquisition system is developed to generate a domain knowledge base that represents a model of the syntactical knowledge that a native English speaker must acquire in order to be proficient at reading technical Japanese. These rules are used as the domain knowledge base against which a student's performance is measured. A student model is designed to assess a student's competence of the Japanese grammar, and to individualize the instruction of an intelligent tutoring system.