Abstract
Understanding a natural language requires knowledge about that language as a system of representation. Further, when the task is one of understanding an extended discourse, world knowledge is also required. This thesis explores some of the issues involved in representing both kinds of knowledge, and also makes an effort to arrive at some under standing of the relationship between the two. A part of this exploration involves an examination of some natural language understanding systems which have attempted to deal with extended discourse both in the form of stories and in the form of dialogues. The systems exam ined are heavily dependent on world knowledge. Another part of this exploration is an effort to build a dialogue system based on speech acts and practical argu ments, as they are described in "Recognizing Promises, Advice, Threats, and Warnings", a Masters Thesis presented to Rochester Institute of Technology, School of Computer Science and Technology, in 1986 by Kevin Donaghy. This dialogue system includes a deterministic syntactic parser, a semantic representation based on the idea of case frames, and a context interpreter that recognizes and represents groups of sentences as practical arguments. This Prolog implementation employs a frame package developed and described in "A Frame Virtual Machine in C-Prolog", a Masters Thesis presented to Rochester Institute of Technology, School of Computer Science and Technology, in 1987 by LeMora Hiss. While this automated dialogue system is necessarily limited in the domain that it recognizes, the opportunity it allows to build a mechanism and a system of representation brings with it a range of issues from the syntactic, through the semantic, to the contextual and the pragmatic. Here, the focus of inquiry came to settle in the semantic representa tion, where the relationship between knowledge about language and knowledge about the world seems to be naturally resident.
Library of Congress Subject Headings
Natural language processing (Computer science); Knowledge acquisition (Expert systems)
Publication Date
1991
Document Type
Thesis
Department, Program, or Center
Computer Science (GCCIS)
Advisor
Not Listed
Recommended Citation
Huber, Bernard J. Jr., "A Knowledge-based approach to understanding natural language" (1991). Thesis. Rochester Institute of Technology. Accessed from
https://repository.rit.edu/theses/591
Campus
RIT – Main Campus
Comments
Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: QA76.9.N38 H82 1991