Sentence extraction and rhetorical classification for flexible abstracts
Simone Teufel, Marc Moens · 1998
Knowledge about he discourse-level structure of a sci-entific article isuseful for flexible and sub-domain in-dependent automatic abstraction. We are interested in the automatic identification of content units ("argu-mentative entities") in the source text, such as GOAL OR PROBLEM STATEMENT, CONCLUSIONS and RE-SULTS. In this paper, we present an extension fKu-piec et al.’s methodology fortrainable statistical sen-tence xtraction (1995). Our extension additionally classifies theextracted s ntenccs a cording to their ar-gumentative status; because only low-level properties of the sentence are taken into account and no exter-nal knowledge sources other than meta-level linguis-tic ones are used, it achieves robustness and partial domain-independence.