Summarizing Similarities and Differences Among Related Documents

  • Authors:
  • Inderjeet Mani;Eric Bloedorn

  • Affiliations:
  • The MITRE Corporation, W640, 11493 Sunset Hills Road Reston, VA 22090, USA. imani@mitre.org;bloedorn@mitre.org

  • Venue:
  • Information Retrieval
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

In many modern information retrieval applications, a common problemwhich arises is the existence of multiple documents covering similarinformation, as in the case of multiple news stories about an event ora sequence of events. A particular challenge for text summarization isto be able to summarize the similarities and differences ininformation content among these documents. The approachdescribed here exploits the results of recent progress in informationextraction to represent salient units of text and their relationships.By exploiting meaningful relations between units based on ananalysis of text cohesion and the context in which thecomparison is desired, the summarizer can pinpoint similarities anddifferences, and align text segments. In evaluation experiments, thesetechniques for exploiting cohesion relations result in summaries which(i) help users more quickly complete a retrieval task (ii) result inimproved alignment accuracy over baselines, and (iii) improveidentification of topic-relevant similarities and differences.