ECAI 2004 Conference Paper

[full paper]

An Argumentative Approach to Assessing Natural Language Usage based on Web Linguistic Corpora

Carlos Chesñevar, Ana Maguitman

Although spelling and grammar checkers used in word-processing software have helped to significantly reduce the overall amount of burden for checking documents, the problem of judging the appropriateness of language usage in different contexts remains to a large extent still unsolved. This paper presents a novel approach to providing proactive assistance for language usage assessment. In the proposed framework, a critiquing system analyzes the textual information presented by the user as an input to a word-processing tool. Textual expressions from the user's document are extracted and used for querying a web search engine to gather information about their absolute and relative frequencies with respect to a Web-based linguistic corpus. Such frequencies are used to compute usage indices, which are good indicators of the suitability of an expression in a given context. The user's preferences consist of a number of (possibly defeasible) rules and facts which encode different aspects of adequate language usage, defining the acceptability of a given expression on the basis of the computed usage indices. A defeasible argumentation system determines if a given expression is acceptable by analyzing the user's preferences and establishing if there exists a warranted argument supporting this claim. Those expressions assessed as unsuitable are further inspected automatically by the system to help the user make the necessary repairs.

Keywords: natural language processing, defeasible argumentation, web-based systems, critiquing systems

Citation: Carlos Chesñevar, Ana Maguitman: An Argumentative Approach to Assessing Natural Language Usage based on Web Linguistic Corpora. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.581-585.