Elamur Research⌕
☰
home / papers / 2007 / A Multidimensional Approach to Utterance Segmentation and Dialogue Act Classification

A Multidimensional Approach to Utterance Segmentation and Dialogue Act Classification

Summary

The paper addresses how to divide dialogue into units for dialogue act labelling when one stretch of speech can serve several communicative purposes at once. The authors argue that classic utterance or turn segmentation cannot capture such cases: units may be discontinuous, span several turns, or overlap. Overlapping units mean that no single segmentation of a turn yields all relevant units. They adopt the notion of a "functional segment" and propose segmenting dialogue separately for each dimension of a multidimensional taxonomy (DIT++), rather than imposing one segmentation on all dimensions. Two studies test this. The first uses Dutch help-desk dialogues from the DIAMOND corpus to compare per-dimension segmentation against a single shared segmentation. The second uses DIAMOND and English AMI meetings to classify communicative functions automatically with three learners: Ripper, Naive Bayes and IB1. Per-dimension segmentation gave better classification results, and all learners clearly beat the baseline, with Ripper performing best. This suggests that functions in the main dimensions can be learned automatically, which supports richer dialogue analysis and more capable dialogue systems.

Key points

  • Data: 952 utterances (1,408 functional segments) from the human-human part of the DIAMOND corpus, and three scenario-based AMI meetings providing 3,676 functional segment instances. Both datasets were annotated with the DIT++ tagset, which covers 11 dimensions.
  • Task was the most frequent dimension in both datasets (45.6% in DIAMOND, 33% in AMI), followed by Auto-Feedback. In AMI, Time Management was also prominent at 20.3%.
  • Ripper outperformed Naive Bayes and IB1 overall. On full functional tags (97 distinct tags, 132 tag combinations), it reached 50.2% accuracy against a 25.7% baseline.
  • Different cues mattered for different dimensions:
    • Task functions relied mainly on bag-of-words features.
    • Questions also relied on prosody, such as pitch variation.
    • Auto-Feedback relied on words like "okay" or "uh-huh" and on dialogue history.
    • Turn and Time Management relied on short duration and soft, less voiced speech.
  • Some dimensions, such as Contact Management and Social Obligation Management, were rare in the AMI data. They scored above 99% but did not beat the baseline. The authors plan to annotate more data and study multimodal input.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Speech act theory

Cited by

10 papers of our library cite it (of the 140673 whose bibliographies we read); 2 of them have a page here, 8 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2007.sigdial-1.26

@inproceedings{geertzen2007multidimensional,
  title = {{A Multidimensional Approach to Utterance Segmentation and Dialogue Act Classification}},
  author = {Geertzen, Jeroen and Petukhova, Volha and Bunt, Harry},
  booktitle = {Proceedings of the 8th SIGdial Workshop on Discourse and Dialogue},
  year = {2007},
  doi = {10.18653/v1/2007.sigdial-1.26},
  url = {https://doi.org/10.18653/v1/2007.sigdial-1.26}
}

References

14 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jens Allwood. 2000. An activity-based approach to pragmatics. In Harry Bunt and William Black, editors, Abduction, Belief and Context in Dialogue; Studies in Computational Pragmatics, pages 47–80. John Benjamins, Amsterdam, The Netherlands. ↗
  2. Harry Bunt and Amanda Schiffrin. 2007. Def ning interoperable concepts for dialogue act annotation. In Proceedings of the Seventh International Workshop on Computational Semantics (IWCS), pages 16–27. find
  3. Harry Bunt. 2006. Dimensions in dialogue annotation. In Proceedings of the 5th International Conference on Language Resources and Evaluation (LREC 2006). find
  4. Alexander Clark. 2003. Machine learning approaches to shallow discourse parsing: A literature review. IM2.MDM Project Deliverable, March. find
  5. William W. Cohen. 1995. Fast effective rule induction. In Proceedings of the 12th International Conference on Machine Learning (ICML’95), pages 115–123. ↗
  6. Mark G. Core and James F. Allen. 1997. Coding dialogues with the DAMSL annotation scheme. In Working Notes: AAAI Fall Symposium on Communicative Action in Humans and Machines, pages 28–35. find
  7. Walter Daelemans, Antal van den Bosch, and Jakub Zavrel. 1999. Forgetting exceptions is harmful in language learning. Machine Learning, 34(1/3):11–43. ↗
  8. Jeroen Geertzen, Yann Girard, and Roser Morante. 2004. The diamond project. Poster at the 8th Workshop on the Semantics and Pragmatics of Dialogue (CATALOG 2004) find
  9. Piroska Lendvai, Antal van den Bosch, and Emiel Krahmer. 2003. Machine learning for shallow interpretation of user utterances in spoken dialogue systems. In Proceedings of EACL-03 Workshop on Dialogue Systems: interaction, adaptation and styles of management, pages 69–78. find
  10. Elmar Nöth, Anton Batliner, Volker Warnke, Johannes-Peter Haas, Manuela Boros, Jan Buckow, Richard Huber, Florian Gallwitz, Matthias Nutt, and Heinrich Niemann. 2002. On the use of prosody in automatic dialogue understanding. Speech Communication, 36(1-2):45–62. ↗
  11. Volha V. Petukhova and Harry Bunt. 2007. A multidimensional approach to multimodal dialogue act annotation. In Proceedings of the Seventh International Workshop on Computational Semantics (IWCS), pages 142–153. find
  12. Elizabeth Shriberg, Rebecca Bates, Andreas Stolcke, Paul Taylor, Daniel Jurafsky, Klaus Ries, Noah Coccaro, Rachel Martin, Marie Meteer, and Carol Van Ess-Dykema. 1998. Can prosody aid the automatic classif cation of dialog acts in conversational speech? Language and Speech (Special Issue on Prosody and Conversation), 41(3-4):439–487. find
  13. Andreas Stolcke, Klaus Ries, Noah Coccaro, Elizabeth Shriberg, Rebecca Bates, Daniel Jurafsky, Paul Taylor, Rachel Martin, Carol Van Ess-Dykema, and Marie Meteer. 2000. Dialogue act modeling for automatic tagging and recognition of conversational speech. Computational Linguistics, 26(3):339–373. ↗ our card
  14. Ian H. Witten and Eibe Frank. 2000. Data mining: practical machine learning tools and techniques with Java implementations. Morgan Kaufmann Publishers, San Francisco:CA, USA. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/geertzen-2007-multidimensional-approach-utterance-segmentation-dialogue/