Abstract
Utterance rewriting condenses a sparse, multi-turn context into a single, self-contained utterance. It has demonstrated superb effectiveness in response generation. Prior techniques mainly rely on machine translation technology to translate the last (incomplete) utterance in the context into a complete utterance. Such a rewriting paradigm contrasts with the main characteristic of the task, that is, the source and target utterances are mostly the same with only a small portion of local edits. Therefore, they rely heavily on large amounts of data to fit the translation model. This paper proposes DuReSE (dialogue utterance rewritten via sequence editing), a neural utterance editor designed for utterance rewriting. DuReSE shapes utterance rewriting as a sentence editing task and then predicts a small set of edit operations for each word in the dialogue context. The model edits the incomplete utterance in two phases where (1) an in-place editor performs word-level editing, and (2) a post-editor then refines the rewritten utterance. Experiments on three benchmark data sets have suggested that DuReSE outperforms baseline models in terms of EM, BLEU, and ROUGE. Specifically, DuReSE outperforms the baselines by up to 14.1% in EM, indicating that it can retrieve more informative and important information, compared with existing techniques.
Similar content being viewed by others
References
Awasthi A, Sarawagi S, Goyal R et al (2019) Parallel iterative edit models for local sequence transduction. In: EMNLP, pp 4251–4261
Bahdanau D, Cho K, Bengio Y (2015) Neural machine translation by jointly learning to align and translate. In: ICLR
Cao Z, Li W, Li S et al (2018) Retrieve, rerank and rewrite: Soft template based neural summarization. In: ACL, pp 152–161
Choi E, He H, Iyyer M et al (2018) Quac: Question answering in context. In: EMNLP, pp 2174–2184
de Barcelos Silva A, Gomes MM, da Costa CA et al (2020) Intelligent personal assistants: a systematic literature review. Expert Syst Appl 147(113):193
Devlin J, Chang MW, Lee K et al (2019) Bert: pre-training of deep bidirectional transformers for language understanding. In: NAACL, pp 4171–4186
Elgohary A, Peskov D, Boyd-Graber J (2019) Can you unpack that? Learning to rewrite questions-in-context. In: EMNLP, pp 5920–5926
Eric M, Krishnan L, Charette F et al (2017) Key-value retrieval networks for task-oriented dialogue. In: SIGDIAL, pp 37–49
Gu J, Lu Z, Li H et al (2016) Incorporating copying mechanism in sequence-to-sequence learning. In: ACL, pp 1631–1640
Joshi M, Levy O, Zettlemoyer L et al (2019) Bert for coreference resolution: baselines and analysis. In: EMNLP, pp 5807–5812
Joshi M, Chen D, Liu Y et al (2020) Spanbert: improving pre-training by representing and predicting spans. Trans Assoc Comput Linguist 8:64–77
Kingma DP, Ba J (2015) Adam: a method for stochastic optimization. In: ICLR
Kumar V, Joshi S (2016) Non-sentential question resolution using sequence to sequence learning. In: COLING, pp 2022–2031
Lee K, He L, Lewis M et al (2017) End-to-end neural coreference resolution. In: EMNLP, pp 188–197
Li J, Jia R, He H et al (2018) Delete, retrieve, generate: a simple approach to sentiment and style transfer. In: ACL, pp 1865–1874
Lin CY (2004) ROUGE: A package for automatic evaluation of summaries. In: Text Summarization Branches Out. ACL, pp 74–81
Liu Q, Chen B, Lou JG et al (2020) Incomplete utterance rewriting as semantic segmentation. In: EMNLP, pp 2846–2857
Liu Y, Ott M, Goyal N et al (2019) Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692
Lu Y, Lin H, Tang J et al (2022) End-to-end neural event coreference resolution. Artif Intell 303(103):632
Malmi E, Krause S, Rothe S et al (2019) Encode, tag, realize: High-precision text editing. In: EMNLP, pp 5057–5068
Ortega D, Vu NT (2017) Neural-based context representation learning for dialog act classification. arXiv preprint arXiv:1708.02561
Pan Z, Bai K, Wang Y et al (2019) Improving open-domain dialogue systems via multi-turn incomplete utterance restoration. In: EMNLP, pp 1824–1833
Panthaplackel S, Allamanis M, Brockschmidt M (2021) Copy that! editing sequences by copying spans. In: The AAAI conference on artificial intelligence, pp 13,622–13,630
Papineni K, Roukos S, Ward T et al (2002) Bleu: a method for automatic evaluation of machine translation. In: ACL, pp 311-318
Pragst L, Rach N, Minker W et al (2018) On the vector representation of utterances in dialogue context. In: LREC
Rastogi P, Gupta A, Chen T et al (2019) Scaling multi-domain dialogue state tracking via query reformulation. In: ACL, pp 97–105
Regan M, Rastogi P, Gupta A et al (2019) A dataset for resolving referring expressions in spoken dialogue via contextual query rewrites (CQR). arXiv preprint arXiv:1903.11783
Ren G, Ni X, Malik M et al (2018) Conversational query understanding using sequence to sequence modeling. In: WWW, pp 1715–1724
Russo C, Madani K, Rinaldi AM (2021) Knowledge acquisition and design using semantics and perception: a case study for autonomous robots. Neural Process Lett 53(5):3153–3168
See A, Liu PJ, Manning CD (2017) Get to the point: summarization with pointer-generator networks. In: ACL, pp 1073–1083
Song L, Xu K, Zhang Y, et al (2020) Zpr2: Joint zero pronoun recovery and resolution using multi-task learning and bert. In: ACL, pp 5429–5434
Su H, Shen X, Zhang R et al (2019) Improving multi-turn dialogue modelling with utterance rewriter. In: ACL, pp 22–31
Vaswani A, Shazeer N, Parmar N et al (2017) Attention is all you need. NIPS 30:5998–6008
Vinyals O, Fortunato M, Jaitly N (2015) Pointer networks. In: NIPS, pp 2692–2700
Wiseman S, Shieber SM, Rush AM (2018) Learning neural templates for text generation. In: EMNLP, pp 3174–3187
Wu W, Wang F, Yuan A et al (2020) CorefQA: coreference resolution as query-based span prediction. In: ACL, pp 6953–6963
Zaidan AA, Zaidan BB (2020) A review on intelligent process for smart home applications based on IoT: coherent taxonomy, motivation, open challenges, and recommendations. Artif Intell Rev 53(1):141–165
Zheng Q, Zhao P, Li Y et al (2021) Spectrum interference-based two-level data augmentation method in deep learning for automatic modulation classification. Neural Comput Appl 33(13):7723–7745
Zheng Q, Zhao P, Zhang D et al (2021) MR-DCAE: Manifold regularization-based deep convolutional autoencoder for unauthorized broadcasting identification. Int J Intell Syst 36(12):7204–7238
Acknowledgements
This research is supported by National Natural Science Foundation of China (Grant No. 62102244, 62032004 and 62272296) and CCF-Tencent Open Research Fund (RAGR20220129).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Jiang, W., Gu, X., Chen, Y. et al. DuReSE: Rewriting Incomplete Utterances via Neural Sequence Editing. Neural Process Lett 55, 8713–8730 (2023). https://doi.org/10.1007/s11063-023-11174-8
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11063-023-11174-8