Elamur Research⌕
☰
home / papers / 2020 / Modeling Local Contexts for Joint Dialogue Act Recognition and Sentiment Classification with Bi-channel Dynamic Convolutions

Modeling Local Contexts for Joint Dialogue Act Recognition and Sentiment Classification with Bi-channel Dynamic Convolutions

Summary

The paper addresses the problem of identifying dialogue acts (DAR) and classifying sentiment (SC) together within a single model. Both are framed as sequence labeling over the utterances of a conversation. Earlier joint systems relied on recurrent encoders such as LSTM. The authors argue that these encoders underuse local context, because nearby utterances in a dialogue carry the most useful cues about acts and sentiment shifts.

Their framework has three layers:

  • A BiLSTM with sinusoidal position embeddings encodes each utterance.
  • Stacked convolutional layers capture dialogue context.
  • Separate linear outputs predict labels for each task.

For the context layers, they first adopt dynamic convolutions (DCN). They then propose a context-aware variant (CDCN), which generates kernels from a window of neighboring utterances rather than from the current input alone. Both are extended to bi-channel versions (BDCN, BCDCN), with one channel per task and latent cross-channel interaction, trained jointly under multi-task learning. DiaBERT supplies dialogue-level pretrained representations.

On the Mastodon and DailyDialog benchmarks, BCDCN reaches state-of-the-art results. This supports the value of explicit local-context modeling for joint dialogue understanding.

Key points

  • Data and code: Experiments use the Mastodon and DailyDialog benchmarks, and the code is released at github.com/ljynlp/BCDCN. Both BDCN and BCDCN beat prior baselines, including DCR-Net, by clear margins, with BCDCN performing best.

  • Why DiaBERT: Standard BERT accepts at most two sentence segments, so the authors use DiaBERT (Liu and Lapata, 2019) to encode whole multi-utterance dialogues.

  • DiaBERT vs. BERT: DiaBERT helps more on the larger DailyDialog set. On small Mastodon, BERT gives the bigger gain. In data-scale tests on DailyDialog, BERT wins when fewer than 20% of training examples are used for fine-tuning.

  • Ablations on DailyDialog: Performance drops when any of these components is removed or replaced:

    • position embeddings;
    • cross-channel interaction;
    • the dynamic kernel;
    • the context-aware convolution (swapped for vanilla, depth-wise, or plain dynamic convolution).

    A fixed 0.5/0.5 loss weighting also performs worse than homoscedastic-uncertainty weighting.

  • Kernel visualizations: Each channel attends to its own task while also drawing on the other channel. BCDCN shows more activated kernel weights than BDCN.

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

8 papers of our library cite it (of the 140673 whose bibliographies we read); 1 of them have a page here, 7 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2020.coling-main.53

@inproceedings{li2020modeling,
  title = {{Modeling Local Contexts for Joint Dialogue Act Recognition and Sentiment Classification with Bi-channel Dynamic Convolutions}},
  author = {Li, Jingye and Fei, Hao and Ji, Donghong},
  booktitle = {Proceedings of the 28th International Conference on Computational Linguistics},
  year = {2020},
  doi = {10.18653/v1/2020.coling-main.53},
  url = {https://doi.org/10.18653/v1/2020.coling-main.53}
}

References

33 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Reinald Kim Amplayo, Jihyeok Kim, Sua Sung, and Seung-won Hwang. 2018. Cold-start aware user and product attention for sentiment classification. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2535–2544. Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. arXiv:1607.06450. 2016. Layer normalization. arXiv preprint Christophe Cerisara, Somayeh Jafaritazehjani, Adedayo Oluokun, and Hoa Le. 2018. Multi-task dialog act and sentiment recognition on mastodon. In Proceedings of COLING, pages 745–754. ↗ our card
  2. Zheqian Chen, Rongqin Yang, Zhou Zhao, Deng Cai, and Xiaofei He. 2018. Dialogue act recognition via crf-attentive structured network. In Proceedings of SIGIR, pages 225–234. find
  3. Pierre Colombo, Emile Chapuis, Matteo Manica, Emmanuel Vignon, Giovanna Varni, and Chloe Clavel. 2020. Guiding attention in sequence-to-sequence models for dialogue act prediction. In Proceedings of AAAI, pages 7594–7601. ↗
  4. Alexis Conneau, Holger Schwenk, Loı̈c Barrault, and Yann Lecun. 2017. Very deep convolutional networks for text classification. In Proceedings of EACL, pages 1107–1116. find
  5. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: pre-training of deep bidirectional transformers for language understanding. In Proceedings of NAACL-HLT, pages 4171–4186. find
  6. Hao Fei, Yafeng Ren, and Donghong Ji. 2019. Implicit objective network for emotion detection. In Proceedings of NLPCC, pages 647–659. find
  7. Hao Fei, Meishan Zhang, and Donghong Ji. 2020a. Cross-lingual semantic role labeling with high-quality translated training corpus. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7014–7026. find
  8. Hao Fei, Yue Zhang, Yafeng Ren, and Donghong Ji. 2020b. Latent emotion memory for multi-label emotion classification. In Proceedings of AAAI, pages 7692–7699. find
  9. Nobuo Inui, Toshiaki Ebe, Bipin Indurkhya, and Yoshiyuki Kotani. 2001. A case-based natural language dialogue system using dialogue act. In Proceedings of SMC, pages 193–198. find
  10. Chen Jia, Xiaobo Liang, and Yue Zhang. 2019. Cross-domain NER using cross-domain language modeling. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2464–2474. find
  11. Lukasz Kaiser, Aidan N Gomez, and Francois Chollet. 2018. Depthwise separable convolutions for neural machine translation. In Proceedings of ICLR. ↗
  12. Nal Kalchbrenner and Phil Blunsom. 2013. Recurrent convolutional neural networks for discourse compositionality. arXiv preprint arXiv:1306.3584. ↗ our card
  13. Nal Kalchbrenner, Edward Grefenstette, and Phil Blunsom. 2014. A convolutional neural network for modelling sentences. In Proceedings of ACL, pages 655–665. ↗ our card
  14. Simon Keizer, Rieks op den Akker, and Anton Nijholt. 2002. Dialogue act recognition with bayesian networks for dutch dialogues. In Proceedings of SIGDIAL Workshop on Discourse and Dialogue, pages 88–94. find our card
  15. Alex Kendall, Yarin Gal, and Roberto Cipolla. 2018. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics. In Proceedings of CVPR, pages 7482–7491. find
  16. Minkyoung Kim and Harksoo Kim. 2018. Integrated neural network model for identifying speech acts, predicators, and sentiments of dialogue utterances. Pattern Recognition Letters., 101:1–5. ↗
  17. Yoon Kim. 2014. Convolutional neural networks for sentence classification. In Proceedings of EMNLP, pages 1746–1751. find our card
  18. Harshit Kumar, Arvind Agarwal, Riddhiman Dasgupta, and Sachindra Joshi. 2018. Dialogue act sequence labeling using hierarchical encoder with CRF. In Proceedings of AAAI, pages 3440–3447. ↗ our card
  19. Tao Lei, Regina Barzilay, and Tommi Jaakkola. 2015. Molding cnns for text: non-linear, non-consecutive convolutions. In Proceedings of EMNLP, pages 1565–1575. find
  20. Piroska Lendvai and Jeroen Geertzen. 2007. Token-based chunking of turn-internal dialogue act sequences. In Proceedings of SIGDIAL, pages 174–181. find our card
  21. Yanran Li, Hui Su, Xiaoyu Shen, Wenjie Li, Ziqiang Cao, and Shuzi Niu. 2017. Dailydialog: A manually labelled multi-turn dialogue dataset. In Proceedings of IJCNLP, pages 986–995. ↗ our card
  22. Yang Liu and Mirella Lapata. 2019. Text summarization with pretrained encoders. In Proceedings of EMNLP-IJCNLP, pages 3728–3738. ↗
  23. Navonil Majumder, Soujanya Poria, Devamanyu Hazarika, Rada Mihalcea, Alexander Gelbukh, and Erik Cambria. 2019. Dialoguernn: An attentive RNN for emotion detection in conversations. In Proceedings of AAAI, pages 6818–6825. ↗
  24. Chao Qiao, Bo Huang, Guocheng Niu, Daren Li, Daxiang Dong, Wei He, Dianhai Yu, and Hua Wu. 2018. A new method of region embedding for text classification. In Proceedings of ICLR. find
  25. Libo Qin, Wanxiang Che, Yangming Li, Minheng Ni, and Ting Liu. 2020. Dcr-net: A deep co-interactive relation network for joint dialog act recognition and sentiment classification. In Proceedings of AAAI, pages 8665–8672. find
  26. Vipul Raheja and Joel Tetreault. 2019. Dialogue act classification with context-aware self-attention. In Proceedings of NAACL-HLT, pages 3727–3733. ↗ our card
  27. Yafeng Ren, Yue Zhang, Meishan Zhang, and Donghong Ji. 2016. Improving twitter sentiment classification using topic-enriched multi-prototype word embeddings. In Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence, pages 3038–3044. find
  28. 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
  29. Dinoj Surendran and Gina-Anne Levow. 2006. Dialog act tagging with support vector machines and hidden markov models. In Proceedings of InterSpeech. find
  30. Maryam Tavafi, Yashar Mehdad, Shafiq Joty, Giuseppe Carenini, and Raymond Ng. 2013. Dialogue act recognition in synchronous and asynchronous conversations. In Proceedings of SIGDIAL, pages 117–121. ↗ our card
  31. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Proceedings of NeurIPS, pages 5998–6008. find
  32. Felix Wu, Angela Fan, Alexei Baevski, Yann N Dauphin, and Michael Auli. 2019. Pay less attention with lightweight and dynamic convolutions. In Proceedings of ICLR. ↗ our card
  33. Xiang Zhang, Junbo Zhao, and Yann LeCun. 2015. Character-level convolutional networks for text classification. In Proceedings of NeurIPS, pages 649–657. ↗ our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/li-2020-modeling-local-contexts-joint-dialogue/