Pith. sign in

REVIEW 4 major objections 3 minor 1 cited by

From Intents to Conversations: Generating Intent-Driven Dialogues with Contrastive Learning for Multi-Turn Classification

T0 review · 4 major / 3 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Chain-of-Intent turns single-turn intent logs into coherent multilingual dialogues, and MINT-CL improves multi-turn intent classification with a contrastive ranking loss.

desk verdict A practical pipeline with a real corpus release, but the headline gains are smaller and shakier than the abstract implies; worth engaging, not worth taking at face value. read the letter →

arxiv 2411.14252 v3 pith:VC2AWZLT submitted 2024-11-21 cs.CL cs.AI

Junhua Liu , Yong Keat Tan , Bin Fu , Kwan Hui Lim This is my paper · ORCID
classification cs.CLcs.AI
keywords multi-turnintentclassificationdialoguegenerationhiddenMarkovmodellargelanguagecontrastivelearningmultilingualdatasete-commercechatbotsself-play
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tackles a chicken-and-egg problem in building chatbot intent classifiers: training a good multi-turn intent classification model needs large, labeled, multilingual dialogues, but such dialogues are expensive to collect. It proposes Chain-of-Intent, which learns the typical flow of user intents from real e-commerce chat logs as a hidden Markov model, then uses a large language model to write the actual user questions and chatbot answers for each sampled intent chain. The result is MINT-E, a generated eight-market, 381-intent dialogue corpus. The paper further proposes MINT-CL, a multi-task objective that adds a response-ranking contrastive loss to intent classification, and reports that generated data improves average multi-turn intent classification accuracy from 61.05% to 61.55% compared with single-turn-only training. A sympathetic reader would care because the pipeline promises a cheaper way to produce domain-specific, multilingual training data for conversational AI.

What carries the argument

The load-bearing object is an LLM-enhanced hidden Markov model. The hidden states are user intents, sampled from a transition matrix learned from chat logs; the observations are user utterances, but instead of an emission distribution the paper uses a prompt that asks an LLM to write an utterance consistent with the sampled intent, an example single-turn question, and the conversation history. This replaces the conditional-independence assumption of classic HMMs with context-dependent generation. The second piece is MINT-CL: a shared encoder feeds both a hierarchical intent-classification head (local label attention plus a HiTIN-style tree network, combined by beam search) and a contrastive response-ranking head that must pull the embedding of a high-quality final answer closer to the conversation than a lower-quality alternative. The combined objective is $\mathcal{L} = \mathcal{L}_{\text{intent}} + 0.3 \mathcal{L}_{\text{contrastive}}$.

What would settle it

Take a sample of real chat logs, have human annotators label the true intent of every turn, build the transition matrix from those labels, and compare it with the matrix estimated by the paper's single-turn labeler, for example by KL divergence or by sampling chains from both and asking humans which are more realistic; if the single-turn-derived matrix is substantially different, the HMM sampling step is not producing authentic multi-turn intent flows.

Watch

Extended reading notes

Core claim

The paper's central claim is that realistic multi-turn dialogues can be manufactured from two ingredients it already has: single-turn question-intent pairs and unlabeled chat logs. It estimates the turn-length distribution, initial intent distribution, and intent-to-intent transition matrix from roughly 100,000 chat sessions per market, samples a chain of intents, and lets an LLM generate each question and answer conditioned on that intent and the conversation history, while a smaller Llama-3 model provides a second candidate final answer that GPT-4 ranks against the primary one. The paper claims the resulting MINT-E dialogues score higher on GPT-4 quality ratings (7.89 average) than real human chat logs (7.25), and that training an XLM-RoBERTa-based hierarchical classifier with multi-task contrastive learning on MINT-E plus single-turn data beats single-turn-only training in average MTIC accuracy. The gains are concentrated in English and higher-resource markets; results in low-resource languages are weaker.

Load-bearing premise

The whole pipeline rests on the assumption that intents inferred by a single-turn model from raw chat logs are accurate enough that, averaged over many sessions, the transition statistics reflect real multi-turn user behavior.

Editorial extensions

If this is right

  • If the pipeline works, a company can build a multi-turn intent training set by reusing its existing single-turn intent labels and chat-log statistics, avoiding per-dialogue human annotation.
  • The released MINT-E corpus gives the research community a multilingual benchmark with 381 intents across eight markets, far broader than the 12 intents in MSDialog.
  • MINT-CL shows that an auxiliary response-ranking contrastive task can transfer representational quality to intent classification, a result that should hold for other encoder-based dialogue understanding tasks.
  • For low-resource languages, the paper's own numbers suggest generated-data gains are smaller, so the method's practical value depends on LLM quality in the target language.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An implication the paper leaves implicit is that the recipe does not depend on e-commerce specifically; any domain with logged customer sessions and single-turn intent labels, such as banking, telecom, or health, could use the same estimate-and-generate loop.
  • The intent transition matrix could serve as a diagnostic: cross-market differences in intent flows might reveal cultural or platform-specific service patterns, and the generated corpus could be re-balanced to emphasize rare intents.
  • A stronger test of the framework would be to measure downstream accuracy against a human-annotated multi-turn corpus of equal size; the paper compares against single-turn training, not against an equal-size human multi-turn training set.
  • The contrastive ranking signal could be combined with human preference data instead of LLM-generated alternatives, potentially aligning the classifier with human notions of helpful responses.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 3 minor

Summary. The paper introduces Chain-of-Intent, a pipeline that combines Hidden Markov Models with LLM-based in-context generation to synthesize intent-aware, multilingual e-commerce dialogues. Domain statistics (turn distribution and intent transition matrix) are extracted from roughly 100,000 real chat sessions per market, then used to sample intent chains; an LLM generates user questions and agent answers conditioned on history. The authors also propose MINT-CL, a multi-task contrastive learning objective for multi-turn intent classification, and release MINT-E, a synthetic multilingual dialogue corpus covering eight markets and 381 intents. The evaluation reports GPT-4-based dialogue quality scores (Table 3) and downstream MTIC accuracy for single-turn (ST), multi-turn (MT), and multi-task variants (RR, CR) in Table 4, with low-resource results in Table 5. The central claims are that the framework improves dialogue generation quality and classification accuracy, 'particularly in multilingual settings.'

Significance. If the empirical claims held, the paper would provide a valuable resource: MINT-E is substantially larger in intent coverage than existing corpora, spans eight markets with diverse languages, and the code/data release is a concrete contribution. The idea of grounding LLM self-play in empirical intent-transition statistics is sensible and potentially reusable. However, the strength of the contribution as currently evidenced is limited: the dialogue-quality result rests entirely on GPT-4 judging GPT-4-generated text, the classification gain on the main table is 0.5 percentage point on average with no significance testing, and the low-resource table shows consistent degradation. The dataset and pipeline are still of interest, but the paper's headline claims need stronger and more carefully reported evidence.

major comments (4)
  1. [Table 5, Section 5.4.3] The Avg column in Table 5 is arithmetically incorrect: it repeats the BR values (65.89% and 61.41%) instead of averaging the three markets. The corrected averages are approximately 71.09% for ST and 63.36% for MT, meaning MT is worse than ST by about 7.7 points on average and is worse in every market shown (BR, TH, VN). This directly contradicts the abstract's claim of gains 'particularly in multilingual settings' and must be addressed, either by corrected reporting and a revised interpretation or by additional evidence that low-resource performance is not systematically harmed.
  2. [Section 5.2, Section 3.2.4, Table 3] The dialogue-quality result (MINT-E 7.89 vs. Golden 7.25 in Table 3) rests entirely on GPT-4 ratings of conversations that were generated by GPT-4. The paper acknowledges possible preference bias, but provides no human evaluation, no inter-annotator agreement, and no alternative quality signal. Separately, the response-ranking pairs used to train MINT-CL (Section 3.2.4) are also labeled by GPT-4 point-wise scoring. This creates a self-referential loop for both the quality and ranking components: the evaluation and the auxiliary training signal may reflect GPT-4's stylistic preferences rather than human-judged dialogue quality. Some human judgments, or at minimum a cross-model and perturbation analysis, are needed before the quality and ranking claims can be considered established.
  3. [Section 5.4.2, Table 4] The headline classification improvement of MT+CR over ST is 0.50 percentage points on average (61.55% vs. 61.05%), with no confidence intervals, significance tests, or multiple-seed results reported. The per-market pattern is also mixed: MT+CR is worse than ST in ID (63.67 vs. 66.67) and TW (61.65 vs. 63.72), and the benefit is concentrated in English-speaking markets. Given the small magnitude and inconsistent direction, the claim that the framework improves classification accuracy is not statistically or evidentially supported as written. The authors should report variance, run significance tests, or reframe the claim to match the actual evidence.
  4. [Section 3.1.2] The intent transition matrix and initial intent distribution are computed from chat-log intents inferred by a single-turn intent recognition model that ignores conversation history. The paper invokes the Law of Large Numbers to argue that statistics become accurate with enough chat logs, but systematic misclassification of context-dependent intents is a bias, not a random error, and will not vanish with sample size. This affects whether the sampled chains of intent reflect real multi-turn user behavior. The authors should analyze the labeler's error on context-dependent cases, or otherwise validate the transition statistics, rather than relying solely on sample size.
minor comments (3)
  1. [Section 3.1.2, Equation 1] The normalization constraints appear as 'P𝑃(𝑇) = 1' and 'P𝑃(𝐼𝑡|𝐼𝑡−1) = 1'; this seems to be a typesetting artifact and should read Σ𝑃(𝑇)=1 and Σ𝑃(𝐼𝑡|𝐼𝑡−1)=1.
  2. [Section 5.1, Table 2] The text states that SG, MY, and PH primarily use English, but Table 2 lists MY as English/Malay and PH as English/Filipino. Please clarify the primary language classification, especially because the 'English-only markets' average in Table 4 uses an implicit market grouping.
  3. [Section 5.4.2, Table 4] The caption of Table 4 says 'averages across all markets,' but only five markets are reported; BR, TH, and VN appear separately in Table 5. The selection criterion for splitting the tables should be explained in the caption or text.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the classification claim is evaluated on manually annotated real chat logs, and the GPT-4 self-evaluation is an acknowledged validity limitation rather than a definitional reduction.

full rationale

The central MTIC claim is grounded externally. Section 5.3 states that the test sets 'were drawn from online chat logs, with the intents of the last questions manually annotated,' so the accuracy numbers in Tables 4-5 are not produced by the generative pipeline being evaluated. The intent transition matrix and turn distribution in Sections 3.1.1-3.1.2 are estimated from roughly 100,000 real chat sessions per market; although per-turn intents are inferred by a model trained on single-turn data D, Section 3.1.2 flags this explicitly, and no reported classification number is constructed from that matrix by definition. The MINT-CL response-ranking pairs are labeled by GPT-4 (Section 3.2.4), but they serve as an auxiliary training signal; the final evaluation of intent classification uses manual labels, so the gain of MT+CR over ST is an empirical outcome, not a fitted parameter renamed as a prediction. The dialogue-quality comparison in Table 3 is a GPT-4 evaluation of dialogues that the same kind of model helped generate; the paper itself acknowledges 'preference bias toward GPT-generated content' in Section 5.2. This is an evaluation-validity limitation and a reviewer-level correctness risk, but it is not a circular derivation: there is no equation or definition in the paper that makes the reported quality or accuracy score identical to the model's own input or output by construction. Self-citations [18]-[20] are contextual and are not load-bearing for the empirical claims. The incorrect Avg entries in Table 5 are an arithmetic consistency issue, not a circularity issue. No circular step is established.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The results depend on empirical inputs (chat log statistics, single-turn labels) whose fidelity is assumed rather than validated, and on GPT-4 serving as both generator and judge. The only explicitly tuned free parameters are alpha and lambda; the transition matrix and turn distribution are estimated from data, not fitted to the target metrics.

free parameters (4)
  • Laplace smoothing alpha = 0.1
    Smoothing parameter for the intent transition matrix (Section 3.1.2). No sensitivity analysis is given, and it directly affects the sampled intent chains.
  • Contrastive loss weight lambda = 0.3
    Tuned from {0.1, 0.3, 0.5} on validation performance (Section 4.3.3). This is a central hyperparameter for MINT-CL.
  • Number of example single-turn questions in generation prompt = 1 to 3
    The question generation prompt in Appendix A.1 includes one to three example enquiries per intent; the exact number is not controlled or ablated.
  • Training hyperparameters = 2e-5 learning rate, batch size 32, 3 epochs
    Fine-tuning choices in Section 5.3.4; standard but required for exact reproduction.
assumptions (5)
  • domain assumption Historical chat log intents inferred by a single-turn model that ignores context approximate the true intent distribution when aggregated (Law of Large Numbers).
    Section 3.1.2 explicitly states the limitation and appeals to LLN; the HMM transition matrix and turn distribution are built from these labels.
  • domain assumption GPT-4 ratings are a valid measure of dialogue quality and response quality.
    Used for dialogue quality evaluation (Section 5.2) and answer ranking for training (Section 3.2.4). The paper acknowledges possible bias toward GPT-generated content but does not validate against human judgments.
  • domain assumption A single-turn question-intent dataset D with accurate labels exists for each market and can serve as emission exemplars.
    Assumed in Section 2.1; the generation pipeline samples example questions from D for each intent.
  • domain assumption Intent transitions are time-homogeneous Markovian.
    Explicit HMM assumption in Section 3.2.1: P(I_t|I_{t-1}) does not depend on t.
  • domain assumption The three-level intent taxonomy is consistent and sufficient across markets.
    MINT-CL predicts intents through this taxonomy (Section 4.2); if the taxonomy misaligns with real user intents, generated labels will be misaligned.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Intents to Conversations: Generating Intent-Driven Dialogues with Contrastive Learning for Multi-Turn Classification." pith.science (2026). https://pith.science/paper/VC2AWZLT

@misc{pith2026241114252,
  author       = {Pith},
  title        = {Pith review of: From Intents to Conversations: Generating Intent-Driven Dialogues with Contrastive Learning for Multi-Turn Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VC2AWZLT}},
  note         = {Machine review of arXiv:2411.14252}
}
read the original abstract

In conversational AI systems, a critical challenge in training effective multi-turn intent classification models lies in the generation of large-scale, domain-specific, multilingual dialogue datasets. In this paper, we introduce Chain-of-Intent, a novel framework that integrates Hidden Markov Models (HMMs) with Large Language Models (LLMs) to generate intent-driven, context-aware dialogues through self-play. Our method first extracts domain-specific intent transition patterns from real-world e-commerce chat logs, which guide the modeling of turn-level dynamics and intent sequences. LLMs are then employed to parameterize the emission probabilities of HMMs, enabling the generation of natural, coherent utterances aligned with predicted intents and dialogue context. We also propose MINT-CL, a multi-task contrastive learning framework for multi-turn intent classification, which improves performance while reducing dependence on large-scale annotated datasets. Empirical results demonstrate that our approach outperforms competitive baselines in dialogue generation quality and classification accuracy, particularly in multilingual settings. To facilitate future research, we release MINT-E, a comprehensive, multilingual, intent-aware multi-turn dialogue corpus derived from the e-commerce domain\footnote{The reproduced source code and dataset are available at https://github.com/junhua/chain-of-intent.

Figures

Figures reproduced from arXiv: 2411.14252 by the authors.

Figure 1
Figure 1. Overview of the proposed pipeline to sample a chain-of-intent and the subsequent intent-aware dialogues generation [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Example of intent-aware multi-turn dialogue and [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of the Chain-of-Intent framework: using [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Deep Learning Approaches for Multimodal Intent Recognition: A Survey

    cs.CL 2025-07 conditional novelty 4.0 of 10

    A survey of deep learning methods for intent recognition, tracing the field from unimodal text, audio, vision, and EEG approaches to multimodal fusion, alignment, knowledge-augmented, and multi-task models.

Reference graph

Works this paper leans on

65 extracted references · 48 canonical work pages · cited by 1 Pith paper

  1. [1]

    Self-seeding and Multi-intent Self-instructing LLMs for Generating Intent-aware Information-Seeking dialogs

    Arian Askari, Roxana Petcu, Chuan Meng, Mohammad Aliannejadi, Amin Abol- ghasemi, E. Kanoulas, and Suzan Verberne. 2024. Self-seeding and Multi-intent Self-instructing LLMs for Generating Intent-aware Information-Seeking dialogs. ArXiv abs/2402.11633 (2024)

    work page Pith review arXiv 2024

  2. [2]

    Hongshen Chen, Zhaochun Ren, Jiliang Tang, Yihong Eric Zhao, and Dawei Yin. 2018. Hierarchical Variational Memory Network for Dialogue Generation. Proceedings of the 2018 World Wide Web Conference (2018)

    work page 2018

  3. [3]

    Meng Chen, Ruixue Liu, Lei Shen, Shaozu Yuan, Jingyan Zhou, Youzheng Wu, Xiaodong He, and Bowen Zhou. 2019. The JDDC Corpus: A Large-Scale Multi- Turn Chinese Dialogue Dataset for E-commerce Customer Service. ArXiv abs/1911.09969 (2019)

    arXiv 2019

  4. [4]

    Heriberto Cuayáhuitl, Steve Renals, Oliver Lemon, and Hiroshi Shimodaira

    work page

  5. [5]

    Yang Deng, Wenxuan Zhang, Wai Lam, Hong Cheng, and Helen Meng. 2022. User satisfaction estimation with sequential dialogue act modeling in goal-oriented conversational systems. In Proceedings of the ACM Web Conference 2022 . 2998– 3008

    work page 2022

  6. [6]

    Ning Ding, Yulin Chen, Bokai Xu, Yujia Qin, Zhi Zheng, Shengding Hu, Zhiyuan Liu, Maosong Sun, and Bowen Zhou. 2023. Enhancing chat language models by scaling high-quality instructional conversations. arXiv preprint arXiv:2305.14233 (2023)

    arXiv 2023

  7. [7]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783 (2024)

    arXiv 2024

  8. [8]

    Yan Fan, Chengyu Wang, Peng He, and Yunhua Hu. 2022. Building Multi-turn Query Interpreters for E-commercial Chatbots with Sparse-to-dense Attentive Modeling. In Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining . 1577–1580

    work page 2022

Show all 65 references
  1. [9]

    Xiang Gao, Yizhe Zhang, Michel Galley, Chris Brockett, and Bill Dolan. 2020. Dialogue Response Ranking Training with Large-Scale Human Feedback Data. ArXiv abs/2009.06978 (2020)

    2020 arXiv

  2. [10]

    Zengguang Hao, Jie Zhang, Binxia Xu, Yafang Wang, Gerard de Melo, and Xiao- long Li. 2023. IntentDial: An Intent Graph based Multi-Turn Dialogue System with Reasoning Path Visualization. ArXiv abs/2310.11818 (2023)

    2023 arXiv

  3. [11]

    Nicholas Kluge Corrêa, Sophia Falk, Shiza Fatimah, Aniket Sen, and Nythamar de Oliveira. 2024. TeenyTinyLlama: open-source tiny language models trained in Brazilian Portuguese. arXiv e-prints (2024), arXiv–2401

    2024

  4. [12]

    Chuyi Kong, Yaxin Fan, Xiang Wan, Feng Jiang, and Benyou Wang. 2024. PlatoLM: Teaching LLMs in Multi-Round Dialogue via a User Simulator. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 7841–7863

    2024

  5. [13]

    Stefan Larson, Anish Mahendran, Joseph J Peper, Christopher Clarke, Andrew Lee, Parker Hill, Jonathan K Kummerfeld, Kevin Leach, Michael A Laurenzano, Lingjia Tang, et al . 2019. An evaluation dataset for intent classification and out-of-scope prediction. ArXiv abs/1909.02027 (2019)

    2019 arXiv

  6. [14]

    Jiwei Li, Michel Galley, Chris Brockett, Jianfeng Gao, and Bill Dolan. 2015. A diversity-promoting objective function for neural conversation models. arXiv preprint arXiv:1510.03055 (2015)

    2015 arXiv

  7. [15]

    Jiwei Li, Will Monroe, Tianlin Shi, Sébastien Jean, Alan Ritter, and Dan Jurafsky

  8. [16]

    Fei Lin, Cong Zhang, Shengqiang Liu, and Hong Ma. 2020. A hierarchical struc- tured multi-head attention network for multi-turn response generation. Ieee Access 8 (2020), 46802–46810

    2020

  9. [17]

    Hsien-chin Lin, Nurul Lubis, Songbo Hu, Carel van Niekerk, Christian Geishauser, Michael Heck, Shutong Feng, and Milica Gašić. 2021. Domain-independent user simulation with transformers for task-oriented dialogue systems. arXiv preprint arXiv:2106.08838 (2021)

    2021 arXiv

  10. [18]

    Junhua Liu and Bin Fu. 2024. Responsible Multilingual Large Language Models: A Survey of Development, Applications, and Societal Impact. arXiv:2410.17532, 2024 (2024)

    2024 arXiv

  11. [19]

    Junhua Liu, Yong Keat Tan, Bin Fu, and Kwan Hui Lim. 2024. Balancing Accu- racy and Efficiency in Multi-Turn Intent Classification for LLM-Powered Dialog Systems in Production. arXiv:2411.12307 (2024)

    2024 arXiv

  12. [20]

    Junhua Liu, Yong Keat Tan, Bin Fu, and Kwan Hui Lim. 2024. LARA: Linguistic- Adaptive Retrieval-Augmentation for Multi-Turn Intent Classification. Proceed- ings of the Empirical Methods in Natural Language Processing (2024)

    2024

  13. [21]

    Ang Lv, Jinpeng Li, Yuhan Chen, Gao Xing, Ji Zhang, and Rui Yan. 2023. DialoGPS: Dialogue Path Sampling in Continuous Semantic Space for Data Augmentation in Multi-Turn Conversations. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (V...

    2023

  14. [22]

    Andrea Madotto, Chien-Sheng Wu, and Pascale Fung. 2018. Mem2seq: Effectively incorporating knowledge bases into end-to-end task-oriented dialog systems. arXiv preprint arXiv:1804.08217 (2018)

    2018 arXiv

  15. [23]

    Do June Min, Paloma Sodhi, and Ramya Ramakrishnan. 2023. Workflow-Guided Response Generation for Task-Oriented Dialogue. ArXiv (2023)

    2023

  16. [24]

    Nikita Moghe, Evgeniia Razumovskaia, Liane Guillou, Ivan Vulić, Anna Korhonen, and Alexandra Birch. 2022. Multi3NLU++: A multilingual, multi-intent, multi- domain dataset for natural language understanding in task-oriented dialogue. arXiv preprint arXiv:2212.10455 (2022)

    2022 arXiv

  17. [25]

    Wenchuan Mu and Kwan Hui Lim. 2022. Revision for Concision: A Constrained Paraphrase Generation Task. InProceedings of the Workshop on Text Simplification, Accessibility, and Readability (TSAR-2022). 57–76

    2022

  18. [26]

    Wenchuan Mu and Kwan Hui Lim. 2023. Modelling Text Similarity: A Survey. In Proceedings of the International Conference on Advances in Social Networks Analysis and Mining. 698–705

    2023

  19. [27]

    Wenchuan Mu and Kwan Hui Lim. 2024. Label-Free Topic-Focused Summariza- tion Using Query Augmentation. In 2024 International Joint Conference on Neural Networks (IJCNN). IEEE, 1–8

    2024

  20. [28]

    Xuan-Phi Nguyen, Wenxuan Zhang, Xin Li, Mahani Aljunied, Qingyu Tan, Liying Cheng, Guanzheng Chen, Yue Deng, Sen Yang, Chaoqun Liu, Hang Zhang, and Li Bing. 2023. SeaLLMs - Large Language Models for Southeast Asia. ArXiv abs/2312.00738 (2023)

    2023 arXiv

  21. [29]

    Jinjie Ni, Tom Young, Vlad Pandelea, Fuzhao Xue, and Erik Cambria. 2023. Recent advances in deep learning based dialogue systems: A systematic survey.Artificial intelligence review 56, 4 (2023), 3055–3155

    2023

  22. [30]

    Oluwatobi Olabiyi, Erik T Mueller, and Rui Zhang. 2023. Multi-turn dialogue re- sponse generation with autoregressive transformer models. US Patent 11,615,255

    2023

  23. [31]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. Advances in neural information processing systems 35...

    2022

  24. [32]

    Bruce Croft, Johanne R

    Chen Qu, Liu Yang, W. Bruce Croft, Johanne R. Trippas, Yongfeng Zhang, and Minghui Qiu. 2018. Analyzing and Characterizing User Intent in Information- seeking Conversations. The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval (2018)

    2018

  25. [33]

    Chen Qu, Liu Yang, W Bruce Croft, Yongfeng Zhang, Johanne R Trippas, and Minghui Qiu. 2019. User intent prediction in information-seeking conversations. In Proceedings of the 2019 Conference on Human Information Interaction and Retrieval. 25–33

    2019

  26. [34]

    Fuji Ren and Siyuan Xue. 2020. Intention detection based on siamese neural network with triplet loss. IEEE Access 8 (2020), 82242–82254

    2020

  27. [35]

    Sebastian Ruder. 2017. An Overview of Multi-Task Learning in Deep Neural Networks. ArXiv abs/1706.05098 (2017)

    2017 arXiv

  28. [36]

    Phillip Rust, Jonas Pfeiffer, Ivan Vulic, Sebastian Ruder, and Iryna Gurevych. 2020. How Good is Your Tokenizer? On the Monolingual Performance of Multilingual Language Models. ArXiv abs/2012.15613 (2020)

    2020 arXiv

  29. [37]

    Sashank Santhanam and Samira Shaikh. 2019. A survey of natural language generation techniques with a focus on dialogue systems-past, present and future directions. arXiv preprint arXiv:1906.00500 (2019)

    2019 arXiv

  30. [38]

    Courville

    Iulian Serban, Tim Klinger, Gerald Tesauro, Kartik Talamadupula, Bowen Zhou, Yoshua Bengio, and Aaron C. Courville. 2016. Multiresolution Recurrent Neural Networks: An Application to Dialogue Response Generation. ArXiv abs/1606.00776 (2016)

    2016 arXiv

  31. [39]

    Courville, and Joelle Pineau

    Iulian Serban, Alessandro Sordoni, Yoshua Bengio, Aaron C. Courville, and Joelle Pineau. 2015. Building End-To-End Dialogue Systems Using Generative Hierar- chical Neural Network Models. In AAAI Conference on Artificial Intelligence

    2015

  32. [40]

    Courville, and Yoshua Bengio

    Iulian Serban, Alessandro Sordoni, Ryan Lowe, Laurent Charlin, Joelle Pineau, Aaron C. Courville, and Yoshua Bengio. 2016. A Hierarchical Latent Variable Encoder-Decoder Model for Generating Dialogues. ArXiv abs/1605.06069 (2016)

    2016 arXiv

  33. [41]

    Zhenqiao Song, Xiaoqing Zheng, Lu Liu, Mu Xu, and Xuan-Jing Huang. 2019. Generating responses with a specific emotion in dialog. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics . 3685–3695

    2019

  34. [42]

    Heydar Soudani, Evangelos Kanoulas, and Faegheh Hasibi. 2023. Data augmenta- tion for conversational ai. InProceedings of the 32nd ACM International Conference on Information and Knowledge Management . 5220–5223

    2023

  35. [43]

    Kanoulas, and Faegheh Hasibi

    Heydar Soudani, Roxana Petcu, E. Kanoulas, and Faegheh Hasibi. 2024. A Survey on Recent Advances in Conversational Data Generation. ArXiv abs/2405.13003 (2024). Generating Intent-Driven Dialogues with Contrastive Learning for Multi-Turn Classification CIKM ’25, November 10–14,...

    2024 arXiv

  36. [44]

    Yuchong Sun, Che Liu, Kun Zhou, Jinwen Huang, Ruihua Song, Xin Zhao, Fuzheng Zhang, Di Zhang, and Kun Gai. 2023. Parrot: Enhancing Multi-Turn Instruction Following for Large Language Models. In Annual Meeting of the Association for Computational Linguistics

    2023

  37. [45]

    Silvia Terragni, Modestas Filipavicius, Nghia Khau, Bruna Guedes, André Manso, and Roland Mathis. 2023. In-context learning user simulators for task-oriented dialog systems. arXiv preprint arXiv:2306.00774 (2023)

    2023 arXiv

  38. [46]

    Thanh Tran, Kai Wei, Weitong Ruan, Ross McGowan, Nathan Susanj, and Grant P Strimel. 2022. Adaptive global-local context fusion for multi-turn spoken lan- guage understanding. In Proceedings of the AAAI Conference on Artificial Intelli- gence, Vol. 36. 12622–12628

    2022

  39. [47]

    Bo-Hsiang Tseng, Yinpei Dai, Florian Kreyssig, and Bill Byrne. 2021. Transferable dialogue systems and user simulators. arXiv preprint arXiv:2107.11904 (2021)

    2021 arXiv

  40. [48]

    Jian Wang, Junhao Liu, Wei Bi, Xiaojiang Liu, Kejing He, Ruifeng Xu, and Min Yang. 2020. Improving knowledge-aware dialogue generation via knowledge base question answering. In Proceedings of the AAAI conference on artificial intelligence, Vol. 34. 9169–9176

    2020

  41. [49]

    Kai Wang, Junfeng Tian, Rui Wang, Xiaojun Quan, and Jianxing Yu. 2020. Multi- Domain Dialogue Acts and Response Co-Generation. In Annual Meeting of the Association for Computational Linguistics

    2020

  42. [50]

    Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. 2022. Self-instruct: Aligning language mod- els with self-generated instructions. arXiv preprint arXiv:2212.10560 (2022)

    2022 arXiv

  43. [51]

    Ting-Wei Wu, Ruolin Su, and Biing-Hwang Juang. 2021. A context-aware hierarchical bert fusion network for multi-turn dialog act detection. ArXiv abs/2109.01267 (2021)

    2021 arXiv

  44. [52]

    Chen Xing, Wei Wu, Yu Wu, Jie Liu, Yalou Huang, Ming Zhou, and Wei-Ying Ma. 2017. Topic aware neural response generation. In Proceedings of the AAAI conference on artificial intelligence , Vol. 31

    2017

  45. [53]

    Puyang Xu and Ruhi Sarikaya. 2014. Contextual domain classification in spoken language understanding systems using recurrent neural network. In 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 136–140

    2014

  46. [54]

    Kai Yang, Xinyu Kong, Yafang Wang, Jie Zhang, and Gerard De Melo. 2020. Reinforcement learning over knowledge graphs for explainable dialogue intent mining. IEEE Access 8 (2020), 85348–85358

    2020

  47. [55]

    Wenhao Yu, Chenguang Zhu, Zaitang Li, Zhiting Hu, Qingyun Wang, Heng Ji, and Meng Jiang. 2022. A survey of knowledge-enhanced text generation.Comput. Surveys 54, 11s (2022), 1–38

    2022

  48. [56]

    Guo, Jun Xu, and Xueqi Cheng

    Hainan Zhang, Yanyan Lan, J. Guo, Jun Xu, and Xueqi Cheng. 2018. Reinforcing Coherence for Sequence to Sequence Model in Dialogue Generation. In Interna- tional Joint Conference on Artificial Intelligence

    2018

  49. [57]

    Guo, Jun Xu, and Xueqi Cheng

    Hainan Zhang, Yanyan Lan, J. Guo, Jun Xu, and Xueqi Cheng. 2018. Tailored Sequence to Sequence Models to Different Conversation Scenarios. In Annual Meeting of the Association for Computational Linguistics

    2018

  50. [58]

    Hainan Zhang, Yanyan Lan, Liang Pang, Hongshen Chen, Zhuoye Ding, and Dawei Yin. 2020. Modeling topical relevance for multi-turn dialogue generation. arXiv preprint arXiv:2009.12735 (2020)

    2020 arXiv

  51. [59]

    Saizheng Zhang. 2018. Personalizing dialogue agents: I have a dog, do you have pets too. arXiv preprint arXiv:1801.07243 (2018)

    2018 arXiv

  52. [60]

    Weinan Zhang, Yiming Cui, Yifa Wang, Qingfu Zhu, Lingzhi Li, Lianqiang Zhou, and Ting Liu. 2018. Context-Sensitive Generation of Open-Domain Conversa- tional Responses. In International Conference on Computational Linguistics

    2018

  53. [61]

    Weixiang Zhao, Yanyan Zhao, Xin Lu, Shilong Wang, Yanpeng Tong, and Bing Qin. 2023. Is ChatGPT equipped with emotional dialogue capabilities? ArXiv abs/2304.09582 (2023)

    2023 arXiv

  54. [62]

    Hao Zhou, Minlie Huang, Tianyang Zhang, Xiaoyan Zhu, and Bing Liu. 2018. Emotional chatting machine: Emotional conversation generation with internal and external memory. In Proceedings of the AAAI conference on artificial intelli- gence, Vol. 32

    2018

  55. [63]

    He Zhu, Chong Zhang, Junjie Huang, Junran Wu, and Ke Xu. 2023. HiTIN: Hierarchy-aware Tree Isomorphism Network for Hierarchical Text Classification. In Annual Meeting of the Association for Computational Linguistics

    2023

  56. [2005]

    IEEE Workshop on Automatic Speech Recognition and Understanding, 2005

    Human-computer dialogue simulation using hidden Markov models. IEEE Workshop on Automatic Speech Recognition and Understanding, 2005. (2005), 290– 295

    2005

  57. [2017]

    Adversarial Learning for Neural Dialogue Generation.ArXiv abs/1701.06547 (2017)

    2017 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.