Detecting Discrepancies and Improving Intelligibility: Two Preliminary Evaluations of RIPTIDES

Michael White, Claire Cardie, Vincent Ng, Kiri Wagstaff, and Daryl McCullough.
Proceedings of the 2001 Document Understanding Conference (DUC), 2001.

Click here for the PDF version.

Abstract

We report on two preliminary evaluations of RIPTIDES, a system that combines information extraction (IE), extraction-based summarization, and natural language generation to support user directed multidocument summarization. We report first on a case study of the system's ability to detect discrepancies in numerical estimates appearing in different new articles at different time points in the evolution of a story using a corpus of more than 100 articles from multiple sources about an earthquake in Central America in January 2001. We then report on how our domain-independent extraction-based summarizer performed on the DUC multidocument task, discussing the extent to which we were able to improve cohesion and organization over the baseline, without unduly sacrificing content relevance.

BibTeX entry

@InProceedings{White+al:01b,
  author = {Michael White and Claire Cardie and Vincent Ng and Kiri Wagstaff and Daryl McCullough},
  title = {Detecting Discrepancies and Improving Intelligibility: Two Preliminary Evaluations of {RIPTIDES}},
  booktitle = {Proceedings of the 2001 Document Understanding Conference},
  year = 2001
}