Construction of Natural Language Sentence Acceptors by a Supervised-Learning Technique
Daniel Coulon, Daniel Kayser · IEEE Transactions on Pattern Analysis and Machine Intelligence · 1979
The problem discussed here is to build automatically an acceptor for natural language sentences, a sample of sentences being given. We solve it in the context of computer-assisted instruction and database interrogation. This paper is focused toward the definition of a learning criterion, i.e., a quality measure on the set of the acceptors which are compatible with the sample.