Summarization

From HLT@INESC-ID

People

L2F is conducting preliminary research in the area of multimodal summarization, namely, in the context of broadcast news summarization. Although they are different domains, the extractive summarization techniques commonly used in broadcast news are currently being applied by the research community to meeting summarization. We intend to explore the use of these techniques in the summarization of meetings in Portuguese as well as specific techniques for meeting summarization (e.g. hot spot detection or attention focus). One PhD thesis is currently ongoing in this area.

Ongoing Projects

  • Broadcast news summarization
  • Using background information to improve speech-to-text summarization
    • Mixed-source multi-document summarization
    • Topic-directed semantic spaces

Selected Publications

  • Ricardo Ribeiro and David Martins de Matos. Using Prior Knowledge to Assess Relevance in Speech Summarization, In IEEE 2008 Workshop on Spoken Language Technology, IEEE, 2008.
  • Ricardo Ribeiro and David Martins de Matos. Mixed-Source Multi-Document Speech-to-Text Summarization, In Coling 2008: Proceedings of the 2nd workshop on Multi-source Multilingual Information Extraction and Summarization, pp. 33-40, Coling 2008 Organizing Committee, 2008.
  • Ricardo Ribeiro, David Martins de Matos, Extractive Summarization of Broadcast News: Comparing Strategies for European Portuguese, In Text, Speech and Dialogue, 10th International Conference, TSD 2007, Springer, vol. 4629, pp. 115-122, 2007.

See Also