iDVS: an interactive multi-document visual summarization system

  • Authors:
  • Yi Zhang;Dingding Wang;Tao Li

  • Affiliations:
  • School of Computer Science, Florida International University, Miami, FL;School of Computer Science, Florida International University, Miami, FL;School of Computer Science, Florida International University, Miami, FL

  • Venue:
  • ECML PKDD'11 Proceedings of the 2011 European conference on Machine learning and knowledge discovery in databases - Volume Part III
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multi-document summarization is a fundamental tool for understanding documents. Given a collection of documents, most of existing multidocument summarization methods automatically generate a static summary for all the users using unsupervised learning techniques such as sentence ranking and clustering. However, these methods almost exclude human from the summarization process. They do not allow for user interaction and do not consider users' feedback which delivers valuable information and can be used as the guidance for summarization. Another limitation is that the generated summaries are displayed in textual format without visual representation. To address the above limitations, in this paper, we develop iDVS, a visualization-enabled multi-document summarization system with users' interaction, to improve the summarization performance using users' feedback and to assist users in document understanding using visualization techniques. In particular, iDVS uses a new semi-supervised document summarization method to dynamically select sentences based on users' interaction. To this regard, iDVS tightly integrates semi-supervised learning with interactive visualization for document summarization. Comprehensive experiments on multi-document summarization using benchmark datasets demonstrate the effectiveness of iDVS, and a user study is conducted to evaluate the users' satisfaction.