자연언어처리에 있어서의 의미의 문제

Abstract

While using computers to analyze linguistic structures and to process linguistic information, we often expect them to be able to talk and act like humans. We particularly expect them to know human languages and to become good interpreters. But their knowledge depends on some prepro-grammed database and their actions, on some representational scheme. Here I mainly discuss what it means for a computer to be able to understand a language and a discourse and, in particular, how it can successfully answer questions and convey appropriate information about an observed discourse. I also discuss how various semantic theories or representational schemes may characterize the role or capacity of a computer in processing linguistic information.