Jump to navigation Jump to search
- (Hirst, 2006) ⇒ Graeme Hirst. (2006). “Views of Text-Meaning in Computational Linguistics: Past, present, and future.” In: Computing, Philosophy, and Cognitive Science; Edited by G. Dodig-Crnkovic and S. Stuart.
Subject Headings: Natural Language Processing
- The move away from this [non-statistical] paradigm came with the growing realization that there were many useful natural-language applications in which some degree of error could be tolerated. These include text classification and document routing, text summarization, and finding answers to questions in a document collection.
- In this paper, I’ll use the word text to denote any complete utterance, short or long. In a computational context, a text could be a non-interactive document, such as a news article, a legal statute, or a memorandum, that a writer or author has produced for other people and which is to undergo some kind of processing by a computer. Or a text could be a natural-language utterance by a user in a spoken or typewritten interactive dialogue with another person or a computer: a turn or set of turns in a conversation.2 The term text-meaning, then, as opposed to mere word-meaning or sentence-meaning, denotes the complete in-context meaning or message of such texts at all levels of interpretation including subtext.
2005-2015: Reclaiming the distinctions
- Rather, in the new NLP applications that are now on the horizon, people will use computers to interpret the natural language of the world, not just to search it for information. Moreover, there will be two types of interpretation possible: interpretation on behalf of the user and interpretation on behalf of the writer.
|2006 ViewsOfTextMeaningInCompLingustics||Graeme Hirst||Views of Text-Meaning in Computational Linguistics: Past, present, and future||http://www.cs.toronto.edu/pub/gh/HIDE-Hirst-ECAP-2006.pdf|