Academic Commons

Presentations (Communicative Events)

Columbia at the Document Understanding Conference 2003

McKeown, Kathleen; Barzilay, Regina; Nenkova, Ani; Schiffman, Barry; Sigelman, Sergey; Schlaiker, Andrew; Blair-Goldensohn, Sasha; Hatzivassiloglou, Vasileios

The Columbia Summarizer for DUC 2003, Task 2, is based on the multi-document summarization system that we developed for DUC 2002 (McKeown et al., 2002). It uses different summarization strategies depending on the type of documents in the input set. Four different strategies are used, one for single events, one for multiple related events, one for biographies and one for discussion of an issue with related events. The summarization strategy encoded in MULTIGEN is used for single event document sets. All other strategies are encoded in DEMS, which uses different parameter settings for different document input types. For Task 4, we adapted an open-ended question answering system that we have been developing as part of our AQUAINT project. The version used in DUC contains only a subset of the different techniques contained in the AQUAINT system given the differences in problem specification.

Files

More About This Work

Academic Units
Computer Science
Publisher
Proceedings of the Document Understanding Workshop (DUC 2003)
Published Here
May 17, 2013
Academic Commons provides global access to research and scholarship produced at Columbia University, Barnard College, Teachers College, Union Theological Seminary and Jewish Theological Seminary. Academic Commons is managed by the Columbia University Libraries.