A Discourse Model for Gist Preservation

This paper describes an approach to gist preservation during automatic summarization whereby the source is a complex information structure which must be “pruned” and organized in such a way as to make it appropriate for textual expression. Based on a discourse model, we propose a process whereby gist is guaranteed at the deep level according to communicative and rhetorical settings. The main function of such a goal-driven summarization model is to map intentions onto coherence relations whilst still observing the semantic dependency indicated by the message source. The discourse model is thus based on an association of intentionality, coherence and semantics, which guides the production of summary message sources that highlight the central proposition of the discourse.