Theoretical Studies in Natural Language Understanding.
Woods,William A · Defense Technical Information Center (DTIC) · 1979
Central to being able to construct powerfully intelligent knowledge-based systems is the development of a representational system that is able to handle open-ended domains of knowledge. Critical to this effort is an understanding of the semantics of such notations, and more generally an understanding of the fundamental notions of 'semantics' and 'meaning'. This report addresses fundamental issues of semantics for computational systems. The question at issue is 'What is it that machines can have that would correspond to the knowledge of meanings that people have and that seem to refer to by the ordinary language term 'meaning'?' The proposed answer is that the notion of truth-conditions can be explicated and made precise by identifying them with a particular kind of abstract procedure and that such procedures can serve as the meaning bearing elements of a theory of semantics suitable for computer implementation. This theory, referred to as 'procedural semantics,' has been the basis of several successful computerized systems and is acquiring increasing interest among philosophers of language. (Author)