BibSonomy :: publication ::

tag user group author concept BibTeX key search:all search:diego_ma

The blue social bookmark and publication sharing system.

(0)

This publication has not been reviewed yet.

rating distribution

average user rating

The average rating is computed over all reviews. However, some of them may be invisible to you due to the visibility setting chosen by the reviewers.

(0.0 of 5.0 based on 0 reviews)

Sentence Fusion for Multidocument News Summarization

by: Regina Barzilay, and Kathleen R. McKeown

In: Computational Linguistics, Vol. 31, Nr. 3 (September 2005) , p. 297--328.

Abstract

A system that can produce informative summaries, highlighting common information found in many online documents, will help Web users to pinpoint information that they need without extensive reading. In this article, we introduce sentence fusion, a novel text-to-text generation technique for synthesizing common information across documents. Sentence fusion involves bottom-up local multisequence alignment to identify phrases conveying similar information and statistical generation to combine common phrases into a sentence. Sentence fusion moves the summarization field from the use of purely extractive methods to the generation of abstracts that contain sentences not found in any of the input documents and can synthesize information across sources.

BibTeX record

@article{Barzilay:2005,
  abstract = {A system that can produce informative summaries, highlighting common information found in many online documents, will help Web users to pinpoint information that they need without extensive reading. In this article, we introduce sentence fusion, a novel text-to-text generation technique for synthesizing common information across documents. Sentence fusion involves bottom-up local multisequence alignment to identify phrases conveying similar information and statistical generation to combine common phrases into a sentence. Sentence fusion moves the summarization field from the use of purely extractive methods to the generation of abstracts that contain sentences not found in any of the input documents and can synthesize information across sources.},
  added-at = {2010-05-28T07:51:03.000+0200},
  author = {Barzilay, Regina and McKeown, Kathleen R.},
  biburl = {http://www.bibsonomy.org/bibtex/2677204ee6eac41e7b1dea8ca8cbbdd67/diego_ma},
  interhash = {a656df5b0a2be86abc9c4efb1b9f1972},
  intrahash = {677204ee6eac41e7b1dea8ca8cbbdd67},
  journal = {Computational Linguistics},
  keywords = {summarisation},
  library = {Mine (March 2010)},
  month = {September},
  number = 3,
  pages = {297--328},
  review = {This is a seminal paper on multi-document summarisation. The domain is news, and the approach follows the following steps:  1. Cluster the sentences by defining a WordNet-based sentence similarity measure. Each cluster is called a theme.  2. Rank the themes and select the top n.  3. Order the themes based on the timestamp of the earliest sentence of each theme.  4. Apply sentence fusion by following these steps:           a. Identify the information based between the sentences in the theme. This is done by aligning the dependency trees and select the tree fragments in common.       b. Select the most important sentence and use its tree as the starting point       c. Expand the tree by attaching nodes from other sentences in the theme       d. Remove the tree fragments that have not been used, always keeping sentence grammaticality       e. Generate a sentence by producing all possible sentences and selecting the one with lowest entropy according to a reference corpus  Several aspects of the sentence fusion method are very interesting and worth exploring for:     - EBM-based summarisation    - Fingerprinting of groups of documents (ARC Linkage with Topedia)    - Graph-based QA},
  timestamp = {2010-05-28T07:51:03.000+0200},
  title = {Sentence Fusion for Multidocument News Summarization},
  volume = 31,
  year = 2005
}

Endnote record