In Natural Languages Processing, there are lots of syntax trees corresponding to a input sentence. If we can choose the correct syntax tree meaningfully, the quality of the processing is improved. Conventionally, we have used selectional restrictions. But semantic categories used in selectional restrictions are so rough that we couldn't choose the correct syntactic structure precisely. This paper shows the method of building the dependency constraint into a context free grammar by subdividing nonterminals according to the meaning of the phrase generated from the nonterminal and by grasping production rules from superordinate-subordinate relation of their meanings in the thesaurus.
|Number of pages||4|
|Journal||Research Reports on Information Science and Electrical Engineering of Kyushu University|
|Publication status||Published - Sep 1 1996|
All Science Journal Classification (ASJC) codes
- Computer Science(all)
- Electrical and Electronic Engineering