Summary: | 碩士 === 國立雲林科技大學 === 資訊工程研究所 === 97 === This thesis proposes a knowledge engineering tool for natural language understanding. This tool supports teaching systems which need natural language understanding, including language translation, math word problem posing, and writing diagnosis. The tool builds a tree structure of knowledge with frame-slot, and describes every knowledge node with template and fire match rule. The relationship between words is described with three template operations: “:” (contiguous)、 “~” (divided)、 “()” (optional). The tool can import ready-made linguistic knowledge like FrameNet, WordNet and HowNet. Besides, the tool provides diagnostic functions to understand incomplete sentences and identify missing parts.
|