Use of a text grammar for generating highlight abstracts of magazine articles

Abstract

Browsing a database of article abstracts is one way to select and buy relevant magazine articles online. Our research contributes to the design and development of text grammars for abstracting texts in unlimited subject domains. We developed a system that parses texts based on the text grammar of a specific text type and that extracts sentences and statements which are relevant for inclusion in the abstracts. The system employs knowledge of the discourse patterns that are typical of news stories. The results are encouraging and demonstrate the importance of discourse structures in text summarisation.

Automatic summarising: The state of the art

“…In practice also, things are different. Moens and Dumortier (2000)'s automatic summaries for magazine articles were designed to prompt magazine purchases. A proper evaluation would check whether sales increased.…”

“…in DUC), with substantial effort in Japanese (see NTCIR) and work on Chinese (Chan et al 2000), Dutch (Moens and Dumortier 2000) and German (Reithinger et al 2000), and both raw Arabic (Douzidia and Lapalme 2204) and automatically-translated Arabic news in DUC 2004. Systems that deploy NLP resources like dictionaries have to respond to language, but statistical systems need not or do it in minimal ways as in stemming.…”

“…repetition (though the distinction is not absolute). News stories have little explicit structure beyond top headings, and this has not usually been exploited though Moens and Dumortier (2000) pick up quotes as a marked form of structure. Teufel (2001) and Teufel and Moens (2002) use citations, which resemble quotes, in technical text.…”

“…Other explicit structures signalled by e.g. headings, include those in legal material Grover et al 2003;Farzinder and Lapalme 2004), magazine articles (Moens and Dumortier 2000), and technical articles (McKeown et al 1998, Elhadad andLapalme 2000, 2002). Web pages, source in Radev et al (2001a), have complex structures with items that may need excluding (Berger and Mittal 2000b).…”

“…McKeown et al (2002) respond to these in Newsblaster. Moens and Dumortier (2000) distinguish opinions and reportage for magazine pieces, Farzinder and Lapalme (2004) legal direction versus legal narrative. Narrative, description and argument are general genres, and McKeown et al (1998) and Teufel (2001) and Teufel and Moens (2002) respond explicitly to argument and description.…”