Elamur Research⌕
☰
home / papers / 2007 / Token-based Chunking of Turn-internal Dialogue Act Sequences

Token-based Chunking of Turn-internal Dialogue Act Sequences

Summary

Most dialogue act (DA) classifiers assume the input is already split into units such as utterances or turns. In reality, finding unit boundaries is a hard problem of its own. Transcription conventions also differ between corpora, which biases how units map to DA labels. Lendvai and Geertzen avoid relying on hand-made boundaries. They treat DA recognition as token-level chunking, similar to phrase chunking in syntax. Every token in a speaker's stretch of speech, up to the next speaker change, gets a DA label plus a marker for whether it begins or continues a chunk.

They compared two sequence learners, conditional random fields (CRF++) and memory-based tagging (MBT), on two very differently transcribed English corpora: Monroe (DAMSL) and MRDA (ICSI meetings). Both learners clearly outperformed a majority-chunk baseline and local classifiers (Naive Bayes, IB1). Their scores were in a similar range across both datasets. Lexical tokens alone proved as informative as richer feature sets. The authors present the method as an out-of-context baseline for studying how context affects DA chunking.

Key points

  • Datasets: Monroe contributes 8 task-oriented disaster-handling dialogues (2,897 utterances, 1,701 turns, 9 DAMSL labels). MRDA covers 75 multi-party meetings (51,452 turns) with 7 grouped labels: B, D, F, Q, S, X, Z.
  • Input sequences: tokens, including disfluencies, were merged up to each speaker change. This yields about 1.5 (Monroe) and 1.8 (MRDA) DA boundaries per sequence.
  • Best result: a chunk F1 of 47 on MRDA, from MBT using tokens only. On the smaller Monroe set, CRF beat MBT on tokens only (macro F1 30 vs 22).
  • Feature sets: adding a broad set of contextual features, or the previous speaker's bag of words, generally did not beat the token-only setup. It helped only for some individual DA types.
  • Tools: CRF++ was run with default settings, and MBT relied on TiMBL 5.2 with a context window of one token.

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); 5 of them have a page here, 5 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2007.sigdial-1.30

@inproceedings{lendvai2007token,
  title = {{Token-based Chunking of Turn-internal Dialogue Act Sequences}},
  author = {Lendvai, Piroska and Geertzen, Jeroen},
  booktitle = {Proceedings of the 8th SIGdial Workshop on Discourse and Dialogue},
  year = {2007},
  doi = {10.18653/v1/2007.sigdial-1.30},
  url = {https://doi.org/10.18653/v1/2007.sigdial-1.30}
}

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. James Allen and Mark Core. 1997. Draft of DAMSL: Dialog act markup in several layers. Unpublished manuscript. find
  2. Walter Daelemans, Jakub Zavrel, Antal van den Bosch, and Ko van der Sloot. 2003. MBT: Memory based tagger, version 2.0, Reference guide. ILK research group technical report series 03-13, Tilburg. find
  3. Walter Daelemans, Jakub Zavrel, Ko van der Sloot, and Antal van den Bosch. 2004. TiMBL: Tilburg Memory Based Learner, version 5.1, Reference guide. Technical Report 04-02, ILK, Tilburg University, Tilburg, The Netherlands. find
  4. Daniel Jurafsky, Elizabeth Shriberg, and Debra Biasca. 1997. Switchboard SWBD-DAMSL shallow-discourse-function annotation coders manual, draft 13. Technical Report 97-02, Institute of Cognitive Science, University of Colorado, USA. find
  5. John D. Lafferty, Andrew McCallum, and Fernando C. N. Pereira. 2001. Conditional random fields: Probabilistic models for segmenting and labeling sequence data. In ICML ’01: Proceedings of the Eighteenth International Conference on Machine Learning, pages 282–289, San Francisco, CA, USA. Morgan Kaufmann Publishers Inc. find
  6. Piroska Lendvai, Antal van den Bosch, and Emiel Krahmer. 2003. Memory-based Disfluency Chunking. In Proceedings of DISS-03, Disfluency in Spontaneous Speech Workshop, pages 63—66. Answ 23 16 17 6 11 10 Stat 33 29 30 39 35 32 IList 26 19 15 13 18 11 ISpk 21 15 12 9 15 8 IReq 23 16 12 18 11 7 Conv 6 32 0 0 0 16 Oth 26 3 2 6 17 14 Disr 14 1 1 16 16 18 Floorgr 40 38 28 39 31 38 Quest 23 20 10 34 26 31 Statem 38 44 38 46 39 42 X 0 0 0 0 0 0 Z 5 21 14 4 4 5 find
  7. Piroska Lendvai and Antal van den Bosch. 2005. Robust ASR lattice representation types in pragma-semantic processing of spoken input. In Proceedings of the AAAI Spoken Language Understanding Workshop, SLU-2005, pages 15–22. find
  8. Martin F. Porter. 1980. An algorithm for suffix stripping. In Program, 3(14), pages 130–137. ↗
  9. Elizabeth Shriberg, Raj Dhillon, Sonali Bhagat, Jeremy Ang, and Hannah Carvey. 2004. The ICSI meeting recorder dialog act (MRDA) corpus. In Proceedings of the 5th SIGdial Workshop on Discourse and Dialogue, pages 97–100. ↗ our card
  10. Amanda J. Stent. 2000. The Monroe corpus. Technical Report TR728/TN99-2, University of Rochester, Rochester, UK. find
  11. Erik F. Tjong Kim Sang and Jorn Veenstra. 1999. Representing text chunks. In Proceedings of the Ninth Conference on European chapter of the Association for Computational Linguistics (EACL), pages 173–179, Morristown, NJ, USA. Association for Computational Linguistics. find
  12. David R. Traum and Peter A. Heeman. 1996. Utterance Units in Spoken Dialogue. In ECAI Workshop on Dialogue Processing in Spoken Language Systems, pages 125–140. find
  13. Volker Warnke, Ralf Kompe, Heinrich Niemann, and Elmar Nöth. 1997. Integrated dialog act segmentation and classification using prosodic features and language models. In Proceedings of the European Conference on Speech Communication and Technology (EUROSPEECH-97), pages 207– 210. find
  14. Matthias Zimmermann, Yang Liu, Elizabeth Shriberg, and Andreas Stolcke. 2005. Toward joint segmentation and classification of dialog acts in multiparty meetings. In Steve Renals and Samy Bengio, editors, MLMI, volume 3869 of Lecture Notes in Computer Science, pages 187–193. Springer. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/lendvai-2007-token-based-chunking-turn-internal/