Pith. sign in

REVIEW 4 major objections 6 minor 127 references

Dialogue Summarization with Emotion Dynamics Using Topic- and Participant-Centric Decomposition

T0 review · 4 major / 6 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read The paper claims that dialogue summaries preserve both content and emotional dynamics best when the dialogue is decomposed by topic and by participant and explicit emotion labels are added.

desk verdict Useful new pipeline for emotion-aware dialogue summarization, but the AMI emotion metrics are contaminated by target leakage; the IEMOCAP results carry the claim. read the letter →

arxiv 2607.14769 v1 pith:XZQQZ3RE submitted 2026-07-16 cs.CL cs.AIcs.MA

Linyun Xiang , Mark Neerincx , Stephanie Tan This is my paper · ORCID
classification cs.CLcs.AIcs.MA
keywords dialoguesummarizationemotiondynamicstrajectoryevaluationtopicdecompositionparticipant-centricchain-of-agentsmultimodalrecognitionsmalllanguagemodels
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 tries to establish that summarization of multi-party dialogue should treat emotion dynamics as a first-class output alongside content. It proposes a hierarchical multi-agent pipeline that first segments a dialogue topically and also collects each participant's utterances, summarizes each view with per-utterance emotion labels, then aggregates the two kinds of summaries into a final dialogue summary. The central empirical claim is that combining topic- and participant-level summaries produces better final summaries than either alone, and that adding explicit emotion labels and a trajectory-preservation instruction yields the best emotion-related scores. Because no standard evaluation exists for emotional flow, the paper introduces trajectory metrics—compressed speaker emotion sequences compared by edit distance, n-gram overlap, Jaccard, and cosine—and applies them on two multimodal dialogue datasets. A sympathetic reader would care because it makes a concrete design choice: two decomposition perspectives plus emotion information, rather than simply larger models, is what preserves emotional dynamics.

What carries the argument

The load-bearing mechanism is a hierarchical chain-of-agents architecture with five agents: a topic segmenter, an utterance-level emotion recognizer, a topic summarizer, a participant summarizer, and a final aggregator. Topics give chronological flow; participant views give speaker-centric arcs; the aggregator merges both. Evaluation machinery: emotion sequences are compressed by removing consecutive repeated states, then compared with normalized Levenshtein distance, n-gram overlap, Jaccard similarity over transitions, and cosine similarity over transition frequencies.

What would settle it

Have independent human annotators label the emotional trajectory expressed in a set of generated summaries, then compare those labels to the model-extracted trajectories. If summaries with top trajectory-metric scores show no better agreement with human judgments than summaries with bottom scores, the claim that the framework preserves emotional dynamics would be falsified.

Watch

Extended reading notes

Core claim

The central claim is that a hierarchical chain-of-agents framework—where one agent segments a dialogue into topics, another assigns utterance-level emotions from multimodal cues, separate agents produce topic-level and participant-level summaries, and a final agent aggregates them—yields summaries that preserve both semantic content and emotional trajectories. The paper's empirical finding is that the combined (topic + participant) decomposition beats either view alone at the final aggregation stage, and that adding explicit emotion labels plus an instruction to preserve trajectories gives the best emotion-related scores. At the intermediate perspective layer, topic summaries dominate, but p

Load-bearing premise

The central claim rests on the assumption that the emotion trajectories used for evaluation are faithful: on the meeting corpus the reference emotions come from an automated model, and the trajectories are extracted from summaries by a language model from the same family, so high scores could reflect system agreement rather than true emotional fidelity.

Editorial extensions

If this is right

  • Dialogue summarization systems that aim to capture how people feel should decompose input by both topic and participant rather than using a single structural view.
  • Giving the summarizer explicit per-utterance emotion labels and instructing it to preserve trajectories improves emotion-related metrics even when content-focused metrics change little.
  • The proposed compressed emotion-trajectory metrics can serve as a reusable evaluation protocol for any dialogue summary that claims to preserve emotional dynamics.
  • Participant-level summaries, though weaker as intermediate outputs, become competitive at the aggregation stage, suggesting speaker-centric information refines emotional arcs during final synthesis.

Reading between the lines

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

  • If the results transfer, a privacy-preserving variant could drop emotion labels and still retain some trajectory fidelity through instruction alone, which matters when emotion annotation is unavailable or sensitive.
  • The trajectory-metric pattern suggests systems should be tuned depending on emotion diversity: local transition structure is easier to preserve when emotional variance is low, while overall arc capture matters more when variance is high.
  • Because the reference emotions on the meeting corpus come from a model and the trajectory extractor is from the same model family, independent human-labeled trajectories would be the strongest test of whether the reported gains are genuine rather than shared-model bias.
  • A natural extension is to fine-tune the small summarizers directly on the trajectory metrics or add a trajectory-reranking step, which could close the gap between the assisted and full-pipeline conditions.
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 / 6 minor

Summary. The paper proposes a hierarchical Chain-of-Agents framework for multimodal dialogue summarization that decomposes a dialogue into topic-level and participant-level summaries, optionally augmented with automatically inferred utterance-level emotion labels and instructions to preserve emotion trajectories. A final aggregation agent combines these intermediate summaries into a global dialogue summary. The authors introduce four emotion-trajectory metrics (normalized Levenshtein, n-gram overlap, Jaccard, and cosine similarity) computed over compressed speaker-wise emotion sequences, and evaluate the framework on AMI and IEMOCAP using LLaMA 3.2 3B and LLaMA 3.1 8B. The central claims are that combining topic and participant information generally improves summarization performance, and that providing both emotion labels and trajectory-preservation instructions (E+/T+) yields the best results, particularly on emotion-related metrics.

Significance. If the empirical results are reliable, the paper offers a useful framework for integrating affective dynamics into dialogue summarization, and the proposed trajectory metrics are a plausible first step toward evaluating emotional fidelity beyond content overlap. The manuscript is also commendable for shipping a complete implementation and reproducibility package, for structuring the ablation study as a clean 2×2 design, and for honestly acknowledging the lack of human evaluation and the reliance on model-generated annotations. However, the central claims about emotion-trajectory preservation are currently weakened by a target-leakage problem in the AMI evaluation, the absence of statistical validation, and the unvalidated LLM-based trajectory extractor. The IEMOCAP results, which use human emotion labels, provide some support for the framework, but the paper's strongest conclusions depend on the contaminated AMI comparisons.

major comments (4)
  1. [§4–§5, Tables 1–4] AMI emotion-trajectory evaluation is contaminated by target leakage. Section 4 states that AMI lacks emotion annotations and that Emotion-LLaMA is used to obtain emotions; Section 5 defines ground-truth emotion sequences for comparison. In E+ conditions, these same Emotion-LLaMA labels are provided as input to the Topic and Participant Summarization Agents (Eqs. 3–4). The LLaMA 3.1 8B extractor then recovers them from summaries. Thus the E+ vs. E− and Combined vs. Topic-only comparisons on AMI emotion columns measure how well the model echoes input labels, not human-perceived emotional fidelity. The paper's conclusion that E+/T+ is best 'particularly for emotion-related metrics' relies heavily on these contaminated AMI results. Re-run the AMI emotion analysis with human-annotated labels or restrict the AMI emotion-trajectory claims; IEMOCAP emotion results are not affected by this leakag
  2. [§6, Tables 1–4] No error bars, confidence intervals, or significance tests are reported for any metric. The observed differences between configurations are often small (e.g., BL 0.065 vs. 0.072 for LLaMA 3.1 8B in Table 4; NGR 0.468 vs. 0.470 on IEMOCAP). Without per-dialogue variance and paired significance tests (e.g., bootstrap or Wilcoxon), the claim that combining topic and participant information 'generally leads to improved performance' is not statistically grounded. Provide per-dialogue scores and significance tests.
  3. [§5] The predicted trajectories are extracted by LLaMA 3.1 8B from the generated summaries, but the extractor is never validated. Appendix B reports only the Emotion-LLaMA confusion matrix on IEMOCAP, not the extractor's accuracy or agreement with human judgments. If the extractor is biased (as Appendix B suggests for the emotion model), the trajectory metrics may reflect extraction artifacts rather than summary quality. Validate the extractor on a subset with human-annotated emotion sequences or report its confusion matrix on both datasets.
  4. [§6, RQ3] RQ3 reports only delta values (Δ=0.111, 0.301, 0.207, 0.133) without the underlying scores, sample sizes, or distributions. The conclusion that 'the model better preserves local transition structure when emotion trajectories exhibit lower variance' is therefore unsupported. Provide the full per-dialogue values, statistical tests, and ideally control for summary length or topic count.
minor comments (6)
  1. [§4] Typo: 'Its does not affect' should read 'It does not affect'.
  2. [§6, RQ1] Typo: 'Perceptive Modeling Layer' should be 'Perspective Modeling Layer'.
  3. [§4 / Appendix E] The term 'GT-assisted' is misleading for AMI, where emotions are Emotion-LLaMA predictions rather than ground truth. Clarify that only topic boundaries are gold on AMI and that emotion labels are model-generated.
  4. [Appendix C] The output heading for the Participant-specific Summarization Agent says 'Topic summarization with emotion trajectory'; it should say 'Participant summarization with emotion trajectory'.
  5. [Figure 5] The caption says 'on IEMOCAP' but the surrounding text attributes the model's neutral/happiness bias to AMI predictions; please harmonize the wording.
  6. [References] Minor formatting inconsistencies (e.g., 'V . I.' spacing) appear in the reference list; otherwise the bibliography is thorough.

Circularity Check

1 steps flagged · score 6.0 of 10

AMI emotion-trajectory metrics are self-referential: the Emotion-LLaMA reference is also the E+ input, so AMI emotion gains are partly constructed.

  1. self definitional [Section 4 (Datasets and Implementation), Eq. (3)-(4), Section 5 (Emotion Trajectory Evaluation), Tables 3-4 (RQ2)]
    "Topic boundaries are annotated in AMI but not in IEMOCAP; and emotions are annotated for IEMOCAP but not for AMI. Therefore, we use off-the-shelf existing models to obtain topic segments (LLaMA 3.1 8B) and emotions (Emotion-LLaMA)(Cheng et al., 2024), for IEMOCAP and AMI, respectively. Evaluation is conducted using LLaMA 3.1 8B to extract speaker-wise emotion trajectories from generated summaries. The model is prompted to classify emotions expressed by each participant, enabling comparison with ground-truth emotion sequences."

    On AMI, the 'ground-truth emotion sequences' used for trajectory evaluation are the outputs of Emotion-LLaMA. These same Emotion-LLaMA labels are fed into the Topic and Participant Summarization Agents (Eq. 3-4), and in the E+ conditions of Table 4 they are explicitly included as utterance-level emotion labels while the model is instructed to preserve emotion trajectories. The trajectory metrics then compare LLaMA 3.1 8B extractions from the generated summaries with those same Emotion-LLaMA labels. A summarizer that echoes the supplied labels can trivially raise LEV, NGR, JAC, and COS on AMI, so the paper's AMI emotion columns—including the claim that E+/T+ is best 'particularly for emotion-related metrics'—measure label copying rather than preservation of human-perceived emotional dynamic

full rationale

The main architectural claim (topic+participant combination and explicit emotion input help) is partly supported by IEMOCAP, where emotion references are human annotations and the E+ inputs are those human labels; that portion is not circular. However, the paper's strongest emotion-specific conclusions are drawn from AMI, where the 'ground-truth' emotion sequences are Emotion-LLaMA outputs (Section 4). Because the same Emotion-LLaMA labels are supplied as input to the summarization agents (Eq. 3-4) and in the E+ conditions (Table 3), the trajectory metrics in Tables 1, 2, and 4 on AMI compare an LLM extraction of summary emotions to the very labels that were fed into the summarizer. This is a self-definitional evaluation loop for the AMI emotion columns, reinforced by Appendix B showing Emotion-LLaMA's bias toward neutral and happiness. Since the claimed benefit 'particularly for emotion-related metrics' leans heavily on the contaminated AMI emotion results, the central evidence is partially circular; but the IEMOCAP results and BLANC content metrics remain independent, so the score is 6 rather than higher.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new physical or conceptual entities; the agents and metrics are software and evaluation constructs. The central claims rely on the listed assumptions about representational adequacy of discrete emotion labels, proxy metrics, and the reliability of the model-generated references.

assumptions (6)
  • domain assumption Emotion dynamics can be faithfully represented as a sequence of discrete emotion labels from the 7-class set E.
    The entire framework and metrics operate on utterance-level discrete emotion labels; this reduction loses intensity, mixed emotions, and continuous affect. Invoked in Section 3, Eqs. (1)-(2).
  • domain assumption BLANC-help is a valid proxy for summary content quality.
    Section 5 uses BLANC-help as the content metric because no ground-truth summaries exist; its validity for dialogue summaries with emotional content is assumed.
  • domain assumption LLaMA 3.1 8B can reliably extract speaker-wise emotion trajectories from generated summaries.
    Section 5 states 'Evaluation is conducted using LLaMA 3.1 8B to extract speaker-wise emotion trajectories from generated summaries.' No validation against human extraction is provided.
  • ad hoc to paper Emotion-LLaMA predictions can serve as ground-truth emotions for AMI.
    Section 4 explains that AMI lacks emotion annotations, so Emotion-LLaMA predictions are used as the reference; Appendix B shows this model is biased toward neutral and happiness, so the reference itself may be biased.
  • domain assumption Topic segmentation can be treated as a modular preprocessing step without dominating downstream summary quality.
    Appendix B reports Pk=0.41 on AMI, indicating imperfect segmentation, yet segmentation quality is not factored into the main results or uncertainty estimates.
  • domain assumption Aggregating topic- and participant-level summaries preserves long-range dependencies and subtle speaker dynamics.
    The conclusion acknowledges 'may introduce information loss during intermediate aggregation stages' and 'can overlook long-range dependencies and subtle speaker dynamics'; this assumption is load-bearing for the claim that the final summary preserves emotion trajectories.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dialogue Summarization with Emotion Dynamics Using Topic- and Participant-Centric Decomposition." pith.science (2026). https://pith.science/paper/XZQQZ3RE

@misc{pith2026260714769,
  author       = {Pith},
  title        = {Pith review of: Dialogue Summarization with Emotion Dynamics Using Topic- and Participant-Centric Decomposition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XZQQZ3RE}},
  note         = {Machine review of arXiv:2607.14769}
}
read the original abstract

Existing text summarization research has focused much on monologic information (e.g., newspaper articles, reports) without accounting for the interaction between speakers or authors. In contrast, dialogues are a rich communication channel where multiple participants conduct back and forth exchanges to construct meaning. We propose a dialogue summarization framework that explicitly models both semantic and emotion dynamics using multimodal dialogue inputs, built on an adapted hierarchical Chain-of-Agents approach. We decompose dialogues from two perspectives: (1) topic segments based on the utterances of all participants, and (2) participant-specific utterance segments. These are used to generate corresponding summaries while incorporating automatically inferred emotions. Topic- and participant-level summaries are aggregated into a dialogue summary capturing semantic content and emotion trajectories. To evaluate beyond content accuracy, we introduce emotion trajectory metrics measuring how well summaries preserve emotional flow. Experiments with small language models on multimodal dialogue datasets show that our framework produces summaries with both semantic and emotion content. Further experiments on explicit emotion label availability highlight the efficacy of our proposed methodology and the opportunities in dialogue analysis using language models.

Figures

Figures reproduced from arXiv: 2607.14769 by the authors.

Figure 1
Figure 1. Overview of the proposed dialogue summarization framework with Multimodal [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Percentage of emotions classes in ground truth and predicted emotion trajectories. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Example frames from AMI [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Example frames from IEMOCAP. Method AMI Pk AMI Wd Random 0.604 0.751 Even 0.513 0.543 TextTiling 0.391 0.410 BiLSTM 0.447 0.473 BERT 0.331 0.333 S-BERT 0.339 0.334 Our (AMI subset) 0.4126 0.4620 [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]
Figure 5
Figure 5. Figure 5: Confusion Matrix of Emotion-LLaMA on IEMOCAP [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

127 extracted references · 19 canonical work pages

  1. [1]

    Emotion in the Workplace: A Reappraisal , volume =

    Ashforth, Blake and Humphrey, Ronald , year =. Emotion in the Workplace: A Reappraisal , volume =. Human Relations - HUM RELAT , doi =

  2. [2]

    LLM aided semi-supervision for efficient Extractive Dialog Summarization

    Mishra, Nishant and Sahu, Gaurav and Calixto, Iacer and Abu-Hanna, Ameen and Laradji, Issam. LLM aided semi-supervision for efficient Extractive Dialog Summarization. Findings of the Association for Computational Linguistics: EMNLP 2023. 2023. doi:10.18653/v1/2023.findings-emnlp.670

    work page doi:10.18653/v1/2023.findings-emnlp.670 2023

  3. [3]

    SAMS um Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarization

    Gliwa, Bogdan and Mochol, Iwona and Biesek, Maciej and Wawer, Aleksander. SAMS um Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarization. Proceedings of the 2nd Workshop on New Frontiers in Summarization. 2019. doi:10.18653/v1/D19-5409

    doi:10.18653/v1/d19-5409 2019

  4. [4]

    Zero-Shot Strategies for Length-Controllable Summarization

    Retkowski, Fabian and Waibel, Alexander. Zero-Shot Strategies for Length-Controllable Summarization. Findings of the Association for Computational Linguistics: NAACL 2025. 2025. doi:10.18653/v1/2025.findings-naacl.34

    work page doi:10.18653/v1/2025.findings-naacl.34 2025

  5. [5]

    U ni S umm and S umm Z oo: Unified Model and Diverse Benchmark for Few-Shot Summarization

    Chen, Yulong and Liu, Yang and Xu, Ruochen and Yang, Ziyi and Zhu, Chenguang and Zeng, Michael and Zhang, Yue. U ni S umm and S umm Z oo: Unified Model and Diverse Benchmark for Few-Shot Summarization. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023. doi:10.18653/v1/2023.acl-long.718

    work page doi:10.18653/v1/2023.acl-long.718 2023

  6. [6]

    Revisiting Zero-Shot Abstractive Summarization in the Era of Large Language Models from the Perspective of Position Bias

    Chhabra, Anshuman and Askari, Hadi and Mohapatra, Prasant. Revisiting Zero-Shot Abstractive Summarization in the Era of Large Language Models from the Perspective of Position Bias. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 2: Short Papers). 2024. d...

    doi:10.18653/v1/2024.naacl-short.1 2024

  7. [7]

    AER-LLM: Ambiguity-aware Emotion Recognition Leveraging Large Language Models , year=

    Hong, Xin and Gong, Yuan and Sethu, Vidhyasaharan and Dang, Ting , booktitle=. AER-LLM: Ambiguity-aware Emotion Recognition Leveraging Large Language Models , year=

  8. [8]

    Dialogue: The complex whole , volume =

    Weigand, Edda , year =. Dialogue: The complex whole , volume =. Language and Dialogue , doi =

Show all 127 references
  1. [9]

    Cognition and Emotion , volume =

    Paul Ekman , title =. Cognition and Emotion , volume =. 1992 , publisher =. doi:10.1080/02699939208411068 , URL =

    1992 doi

  2. [10]

    A Discourse-Aware Attention Model for Abstractive Summarization of Long Documents

    Cohan, Arman and Dernoncourt, Franck and Kim, Doo Soon and Bui, Trung and Kim, Seokhwan and Chang, Walter and Goharian, Nazli. A Discourse-Aware Attention Model for Abstractive Summarization of Long Documents. Proceedings of the 2018 Conference of the North A merican Chapter o...

    2018 doi

  3. [11]

    arXiv e-prints , keywords =

    Emotion Recognition in Conversation: Research Challenges, Datasets, and Recent Advances. arXiv e-prints , keywords =. doi:10.48550/arXiv.1905.02947 , archivePrefix =. 1905.02947 , primaryClass =

    1905 doi arXiv

  4. [12]

    The Ripple Effect: Emotional Contagion In Groups , volume =

    Barsade, Sigal , year =. The Ripple Effect: Emotional Contagion In Groups , volume =. Yale School of Management, Yale School of Management Working Papers , doi =

  5. [13]

    British Journal of Psychology , volume =

    Emotions Are Social , author =. British Journal of Psychology , volume =

  6. [14]

    , year =

    Sebe, Nicu and Cohen, Ira and Gevers, T. , year =. Multimodal approaches for emotion recognition: A survey , volume =. Proceedings of SPIE - The International Society for Optical Engineering , doi =

  7. [15]

    1971 , address =

    Mehrabian, Albert , title =. 1971 , address =

    1971

  8. [16]

    and Dorris, Alana D

    Ashkanasy, Neal M. and Dorris, Alana D. , title =. Annual Review of Organizational Psychology and Organizational Behavior , volume =. 2017 , doi =

    2017

  9. [17]

    Emotions in meeting talk , isbn =

    Nikander, Pirjo , year =. Emotions in meeting talk , isbn =

  10. [18]

    , title =

    Russell, James A. , title =. Psychological Review , year =

  11. [19]

    Perspectives on Psychological Science , year=

    Are Emotions Natural Kinds? , author=. Perspectives on Psychological Science , year=

  12. [20]

    Emotional Dynamics and Strategizing Processes: A Study of Strategic Conversations in Top Team Meetings , volume =

    Liu, Feng and Maitlis, Sally , year =. Emotional Dynamics and Strategizing Processes: A Study of Strategic Conversations in Top Team Meetings , volume =. Journal of Management Studies , doi =

  13. [21]

    Current Psychology , year =

    Mehrabian, Albert , title =. Current Psychology , year =. doi:10.1007/BF02686918 , issn =

    doi

  14. [22]

    A survey of dialogic emotion analysis: Developments, approaches and perspectives , journal =

    Chenquan Gan and Jiahao Zheng and Qingyi Zhu and Yang Cao and Ye Zhu , keywords =. A survey of dialogic emotion analysis: Developments, approaches and perspectives , journal =. 2024 , issn =. doi:https://doi.org/10.1016/j.patcog.2024.110794 , url =

    2024

  15. [23]

    Trends in cognitive sciences , volume=

    Language as context for the perception of emotion , author=. Trends in cognitive sciences , volume=. 2007 , publisher=

    2007

  16. [24]

    2019 8th International Conference on Affective Computing and Intelligent Interaction (ACII) , year=

    Context in Human Emotion Perception for Automatic Affect Detection: A Survey of Audiovisual Databases , author=. 2019 8th International Conference on Affective Computing and Intelligent Interaction (ACII) , year=

    2019

  17. [25]

    Language and Speech , year =

    Is there an ironic tone of voice? , author =. Language and Speech , year =

  18. [26]

    2022 , eprint=

    Meeting Summarization: A Survey of the State of the Art , author=. 2022 , eprint=

    2022

  19. [27]

    Proceedings of INTERSPEECH , year =

    Extractive Summarization of Meeting Recordings , author =. Proceedings of INTERSPEECH , year =

  20. [28]

    Dialogue Summarization with Mixture of Experts based on Large Language Models

    Tian, Yuanhe and Xia, Fei and Song, Yan. Dialogue Summarization with Mixture of Experts based on Large Language Models. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024. doi:10.18653/v1/2024.acl-long.385

    2024 doi

  21. [29]

    DialogueMLLM: Transforming Multimodal Emotion Recognition in Conversation Through Instruction-Tuned MLLM , volume =

    Sun, Yuanyuan and Zhou, Ting , year =. DialogueMLLM: Transforming Multimodal Emotion Recognition in Conversation Through Instruction-Tuned MLLM , volume =. IEEE Access , doi =

  22. [30]

    Proceedings of the National Academy of Sciences of the United States of America , volume =

    Cross-cultural recognition of basic emotions through nonverbal emotional vocalizations , author =. Proceedings of the National Academy of Sciences of the United States of America , volume =. 2010 , doi =

    2010

  23. [31]

    Carletta, Jean and Ashby, Simone and Bourban, S. The. Machine Learning for Multimodal Interaction , series =. 2005 , doi =

    2005

  24. [32]

    Oxford Research Encyclopedia of Psychology , year =

    Hess, Ursula , title =. Oxford Research Encyclopedia of Psychology , year =

  25. [33]

    Unsupervised Abstractive Meeting Summarization with Multi-Sentence Compression and Budgeted Submodular Maximization

    Shang, Guokan and Ding, Wensi and Zhang, Zekun and Tixier, Antoine and Meladianos, Polykarpos and Vazirgiannis, Michalis and Lorr \'e , Jean-Pierre. Unsupervised Abstractive Meeting Summarization with Multi-Sentence Compression and Budgeted Submodular Maximization. Proceedings...

    2018 doi

  26. [34]

    , author=

    Meetings and more meetings: The relationship between meeting load and the daily well-being of employees. , author=. Group Dynamics: Theory, Research, and Practice , volume=. 2005 , publisher=

    2005

  27. [35]

    , journal=

    Le Ngo, Anh Cat and See, John and Phan, Raphael C.-W. , journal=. Sparsity in Dynamics of Spontaneous Subtle Emotions: Analysis and Application , year=

  28. [36]

    McCowan and J

    I. McCowan and J. Carletta and W. Kraaij and S. Ashby and S. Bourban and M. Flynn and M. Guillemot and T. Hain and J. Kadlec and V. Karaiskos and M. Kronenthal and G. Lathoud and M. Lincoln and A. Lisowska and W. Post and Dennis Reidsma and P. Wellner. The AMI meeting corpus. ...

    2005

  29. [37]

    Mroz and Joseph A

    Joseph E. Mroz and Joseph A. Allen and Dana C. Verhoeven and Marissa L. Shuffler , title =. Current Directions in Psychological Science , volume =. 2018 , doi =. https://doi.org/10.1177/0963721418776307 , abstract =

    2018 doi

  30. [38]

    The Journal of Business Communication (1973) , volume =

    Birte Asmuß and Jan Svennevig , title =. The Journal of Business Communication (1973) , volume =. 2009 , doi =

    1973

  31. [39]

    Emotion Review , volume =

    Michael Boiger and Batja Mesquita , title =. Emotion Review , volume =. 2012 , doi =. https://doi.org/10.1177/1754073912439765 , abstract =

    2012 doi

  32. [40]

    Predicting Meeting Success With Nuanced Emotions , year=

    Zhou, Ke and Constantinides, Marios and Joglekar, Sagar and Quercia, Daniele , journal=. Predicting Meeting Success With Nuanced Emotions , year=

  33. [41]

    Pollmann and Carla A

    Monique M.H. Pollmann and Carla A. Roos , keywords =. “I get u”. People correctly interpret the tone of text messages and emails , journal =. 2025 , issn =. doi:https://doi.org/10.1016/j.chbr.2025.100689 , url =

    2025

  34. [42]

    Wiebe and Rebecca F

    Janyce M. Wiebe and Rebecca F. Bruce and Thomas P. O'Hara , title =. Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics , pages =. 1999 , address =

    1999

  35. [43]

    Paul , title=

    Grice, H. Paul , title=. Syntax and Semantics 3: Speech Acts , editor=. 1975 , pages=

    1975

  36. [44]

    Austin, J. L. , title=. 1962 , publisher=

    1962

  37. [45]

    , title=

    Searle, John R. , title=. 1969 , publisher=

    1969

  38. [46]

    Discourse & Society , volume=

    Goodwin, Charles , title=. Discourse & Society , volume=. 2007 , publisher=

    2007

  39. [47]

    Social IQ a: Commonsense Reasoning about Social Interactions

    Sap, Maarten and Rashkin, Hannah and Chen, Derek and Le Bras, Ronan and Choi, Yejin. Social IQ a: Commonsense Reasoning about Social Interactions. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference ...

    2019 doi

  40. [48]

    , title=

    Gumperz, John J. , title=. 1982 , publisher=

    1982

  41. [49]

    Keep Meeting Summaries on Topic: Abstractive Multi-Modal Meeting Summarization

    Li, Manling and Zhang, Lingyu and Ji, Heng and Radke, Richard J. Keep Meeting Summaries on Topic: Abstractive Multi-Modal Meeting Summarization. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019. doi:10.18653/v1/P19-1210

    2019 doi

  42. [50]

    and Zakariah, Mohammed and Almazyad, Abdulaziz , journal=

    Chandraumakantham, Omkumar and Gowtham, N. and Zakariah, Mohammed and Almazyad, Abdulaziz , journal=. Multimodal Emotion Recognition Using Feature Fusion: An LLM-Based Approach , year=

  43. [51]

    Current Directions in Psychological Science , volume =

    Lisa Feldman Barrett and Batja Mesquita and Maria Gendron , title =. Current Directions in Psychological Science , volume =. 2011 , doi =. https://doi.org/10.1177/0963721411422522 , abstract =

    2011 doi

  44. [52]

    arXiv e-prints , keywords =

    Emotionally-Aware Agents for Dispute Resolution. arXiv e-prints , keywords =. doi:10.48550/arXiv.2509.04465 , archivePrefix =. 2509.04465 , primaryClass =

    doi arXiv

  45. [53]

    How did we get here? Summarizing conversation dynamics

    Hua, Yilun and Chernogor, Nicholas and Gu, Yuzhe and Jeong, Seoyeon and Luo, Miranda and Danescu-Niculescu-Mizil, Cristian. How did we get here? Summarizing conversation dynamics. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computati...

    2024 doi

  46. [54]

    Dialogue acts enhanced extract–abstract framework for meeting summarization , journal =

    Shichao Sun and Ruifeng Yuan and Wenjie Li and Ziqiang Cao and Sujian Li , keywords =. Dialogue acts enhanced extract–abstract framework for meeting summarization , journal =. 2024 , issn =. doi:https://doi.org/10.1016/j.ipm.2023.103635 , url =

    2024

  47. [55]

    2024 , eprint=

    Leveraging Discourse Structure for Extractive Meeting Summarization , author=. 2024 , eprint=

    2024

  48. [56]

    Extractive summarization of multi-party meetings through discourse segmentation , volume =

    Bokaei, Hadi and Sameti, Hossein and Liu, Yang , year =. Extractive summarization of multi-party meetings through discourse segmentation , volume =. Natural Language Engineering , doi =

  49. [57]

    A topic modeling based approach to novel document automatic summarization , journal =

    Zongda Wu and Li Lei and Guiling Li and Hui Huang and Chengren Zheng and Enhong Chen and Guandong Xu , keywords =. A topic modeling based approach to novel document automatic summarization , journal =. 2017 , issn =. doi:https://doi.org/10.1016/j.eswa.2017.04.054 , url =

    2017 doi

  50. [58]

    and Baron, D

    Janin, A. and Baron, D. and Edwards, J. and Ellis, D. and Gelbart, D. and Morgan, N. and Peskin, B. and Pfau, T. and Shriberg, E. and Stolcke, A. and Wooters, C. , booktitle=. The ICSI Meeting Corpus , year=

  51. [59]

    ClusterRank: A Graph Based Method for Meeting Summarization , journal =

    Garg, Nikhil and Favre, Benoit and Riedhammer, Korbinian and Hakkani-Tur, Dilek , year =. ClusterRank: A Graph Based Method for Meeting Summarization , journal =

  52. [60]

    Hierarchical Attention Adapter for Abstractive Dialogue Summarization

    Li, Raymond and Li, Chuyuan and Murray, Gabriel and Carenini, Giuseppe. Hierarchical Attention Adapter for Abstractive Dialogue Summarization. Proceedings of The 5th New Frontiers in Summarization Workshop. 2025. doi:10.18653/v1/2025.newsum-main.2

    2025 doi

  53. [61]

    ACL , year=

    Abstractive Dialogue Summarization with Attentive Hierarchical Transformers , author=. ACL , year=

  54. [62]

    D ialog S um: A Real-Life Scenario Dialogue Summarization Dataset

    Chen, Yulong and Liu, Yang and Chen, Liang and Zhang, Yue. D ialog S um: A Real-Life Scenario Dialogue Summarization Dataset. Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021. 2021. doi:10.18653/v1/2021.findings-acl.449

    2021 doi

  55. [63]

    NAACL , year=

    How Did We Get Here? Summarizing Conversation Dynamics , author=. NAACL , year=

  56. [64]

    EMNLP , year=

    TopicFlow: Modeling Topic Transitions in Dialogue , author=. EMNLP , year=

  57. [65]

    2023 , eprint=

    AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation , author=. 2023 , eprint=

    2023

  58. [66]

    2023 , eprint=

    AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors , author=. 2023 , eprint=

    2023

  59. [67]

    Topic-Oriented Dialogue Summarization , year=

    Lin, Haitao and Zhu, Junnan and Xiang, Lu and Zhai, Feifei and Zhou, Yu and Zhang, Jiajun and Zong, Chengqing , journal=. Topic-Oriented Dialogue Summarization , year=

  60. [68]

    Topic-Aware Contrastive Learning for Abstractive Dialogue Summarization

    Liu, Junpeng and Zou, Yanyan and Zhang, Hainan and Chen, Hongshen and Ding, Zhuoye and Yuan, Caixia and Wang, Xiaojie. Topic-Aware Contrastive Learning for Abstractive Dialogue Summarization. Findings of the Association for Computational Linguistics: EMNLP 2021. 2021. doi:10.1...

    2021 doi

  61. [69]

    Enhancing Dialogue Summarization with Topic-Aware Global- and Local- Level Centrality

    Liang, Xinnian and Wu, Shuangzhi and Cui, Chenhao and Bai, Jiaqi and Bian, Chao and Li, Zhoujun. Enhancing Dialogue Summarization with Topic-Aware Global- and Local- Level Centrality. Proceedings of the 17th Conference of the European Chapter of the Association for Computation...

    2023 doi

  62. [70]

    2022 , eprint=

    A Survey on Dialogue Summarization: Recent Advances and New Frontiers , author=. 2022 , eprint=

    2022

  63. [71]

    M edia S um: A Large-scale Media Interview Dataset for Dialogue Summarization

    Zhu, Chenguang and Liu, Yang and Mei, Jie and Zeng, Michael. M edia S um: A Large-scale Media Interview Dataset for Dialogue Summarization. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technol...

    2021 doi

  64. [72]

    Structure-Aware Abstractive Conversation Summarization via Discourse and Action Graphs

    Chen, Jiaao and Yang, Diyi. Structure-Aware Abstractive Conversation Summarization via Discourse and Action Graphs. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021. doi:10.1865...

    2021 doi

  65. [73]

    2024 , eprint=

    PSentScore: Evaluating Sentiment Polarity in Dialogue Summarization , author=. 2024 , eprint=

    2024

  66. [74]

    QMS um: A New Benchmark for Query-based Multi-domain Meeting Summarization

    Zhong, Ming and Yin, Da and Yu, Tao and Zaidi, Ahmad and Mutuma, Mutethia and Jha, Rahul and Awadallah, Ahmed Hassan and Celikyilmaz, Asli and Liu, Yang and Qiu, Xipeng and Radev, Dragomir. QMS um: A New Benchmark for Query-based Multi-domain Meeting Summarization. Proceedings...

    2021 doi

  67. [75]

    2024 , eprint=

    A Systematic Survey of Text Summarization: From Statistical Methods to Large Language Models , author=. 2024 , eprint=

    2024

  68. [76]

    2024 , issue_date =

    Shakil, Hassan and Farooq, Ahmad and Kalita, Jugal , title =. 2024 , issue_date =. doi:10.1016/j.neucom.2024.128255 , journal =

    2024

  69. [77]

    Long story short – Global unsupervised models for keyphrase based meeting summarization , journal =

    Korbinian Riedhammer and Benoit Favre and Dilek Hakkani-Tür , keywords =. Long story short – Global unsupervised models for keyphrase based meeting summarization , journal =. 2010 , issn =. doi:https://doi.org/10.1016/j.specom.2010.06.002 , url =

    2010 doi

  70. [78]

    A Skip-Chain Conditional Random Field for Ranking Meeting Utterances by Importance

    Galley, Michel. A Skip-Chain Conditional Random Field for Ranking Meeting Utterances by Importance. Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing. 2006

    2006

  71. [79]

    2018 ,URL =

    BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding ,author =. 2018 ,URL =

    2018

  72. [80]

    2023 , eprint=

    Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer , author=. 2023 , eprint=

    2023

  73. [81]

    Tell me what I need to know: Exploring LLM-based (Personalized) Abstractive Multi-Source Meeting Summarization , url=

    Kirstein, Frederic and Ruas, Terry and Kratel, Robert and Gipp, Bela , year=. Tell me what I need to know: Exploring LLM-based (Personalized) Abstractive Multi-Source Meeting Summarization , url=. doi:10.18653/v1/2024.emnlp-industry.69 , booktitle=

    2024 doi

  74. [82]

    ECLM : Entity Level Language Model for Spoken Language Understanding with Chain of Intent

    Yin, Shangjian and Huang, Peijie and Chen, JiaTian and Huang, Haojing and Xu, Yuhong. ECLM : Entity Level Language Model for Spoken Language Understanding with Chain of Intent. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: L...

    2025 doi

  75. [83]

    2025 , eprint=

    Towards Online Multi-Modal Social Interaction Understanding , author=. 2025 , eprint=

    2025

  76. [84]

    2024 , eprint=

    PanoSent: A Panoptic Sextuple Extraction Benchmark for Multimodal Conversational Aspect-based Sentiment Analysis , author=. 2024 , eprint=

    2024

  77. [85]

    2025 , eprint=

    Multi-speaker Attention Alignment for Multimodal Social Interaction , author=. 2025 , eprint=

    2025

  78. [86]

    Towards an ISO Standard for Dialogue Act Annotation

    Bunt, Harry and Alexandersson, Jan and Carletta, Jean and Choe, Jae-Woong and Fang, Alex Chengyu and Hasida, Koiti and Lee, Kiyong and Petukhova, Volha and Popescu-Belis, Andrei and Romary, Laurent and Soria, Claudia and Traum, David. Towards an ISO Standard for Dialogue Act A...

    2010

  79. [87]

    Schegloff and Gail Jefferson

    Harvey Sacks, Emanuel A. Schegloff and Gail Jefferson. A simplest systematics for the organization of turn-taking for conversation. Language. 1974

    1974

  80. [88]

    D ialogue GCN : A Graph Convolutional Neural Network for Emotion Recognition in Conversation

    Ghosal, Deepanway and Majumder, Navonil and Poria, Soujanya and Chhaya, Niyati and Gelbukh, Alexander. D ialogue GCN : A Graph Convolutional Neural Network for Emotion Recognition in Conversation. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Proc...

    2019 doi

  81. [89]

    2019 , eprint=

    DialogueRNN: An Attentive RNN for Emotion Detection in Conversations , author=. 2019 , eprint=

    2019

  82. [90]

    2025 , eprint=

    Omni-Emotion: Extending Video MLLM with Detailed Face and Audio Modeling for Multimodal Emotion Analysis , author=. 2025 , eprint=

    2025

  83. [91]

    DialogueMLLM: Transforming Multimodal Emotion Recognition in Conversation Through Instruction-Tuned MLLM , year =

    Yuanyuan Sun and Ting Zhou , journal =. DialogueMLLM: Transforming Multimodal Emotion Recognition in Conversation Through Instruction-Tuned MLLM , year =

  84. [92]

    Social signal processing: Survey of an emerging domain , journal =

    Alessandro Vinciarelli and Maja Pantic and Hervé Bourlard , keywords =. Social signal processing: Survey of an emerging domain , journal =. 2009 , note =. doi:https://doi.org/10.1016/j.imavis.2008.11.007 , url =

    2009 doi

  85. [93]

    2000 , publisher=

    Affective computing , author=. 2000 , publisher=

    2000

  86. [94]

    COMPARATIVE ANALYSIS OF LARGE LANGUAGE MODELS FOR THE APPLICATION OF SCIENTIFIC ARTICLE SUMMARIZATION , journal =

    Rishabh Saxena and Shubhangi Singh and Preeti Dubey , keywords =. COMPARATIVE ANALYSIS OF LARGE LANGUAGE MODELS FOR THE APPLICATION OF SCIENTIFIC ARTICLE SUMMARIZATION , journal =. 2025 , note =. doi:https://doi.org/10.1016/j.procs.2025.04.002 , url =

    2025 doi

  87. [95]

    Assessing the Impact of Prompt Strategies on Text Summarization with Large Language Models

    Onan, Aytu g and Alhumyani, Hesham. Assessing the Impact of Prompt Strategies on Text Summarization with Large Language Models. Computer Applications in Industry and Engineering. 2025

    2025

  88. [96]

    1986 , publisher=

    The Emotions , author=. 1986 , publisher=

    1986

  89. [97]

    Conversational Memory Network for Emotion Recognition in Dyadic Dialogue Videos

    Hazarika, Devamanyu and Poria, Soujanya and Zadeh, Amir and Cambria, Erik and Morency, Louis-Philippe and Zimmermann, Roger. Conversational Memory Network for Emotion Recognition in Dyadic Dialogue Videos. Proceedings of the 2018 Conference of the North A merican Chapter of th...

    2018 doi

  90. [98]

    Zico and Morency, Louis-Philippe and Salakhutdinov, Ruslan

    Tsai, Yao-Hung Hubert and Bai, Shaojie and Liang, Paul Pu and Kolter, J. Zico and Morency, Louis-Philippe and Salakhutdinov, Ruslan. Multimodal Transformer for Unaligned Multimodal Language Sequences. Proceedings of the 57th Annual Meeting of the Association for Computational ...

    2019 doi

  91. [99]

    arXiv e-prints , keywords =

    A Similarity Measure for Comparing Conversational Dynamics. arXiv e-prints , keywords =. doi:10.48550/arXiv.2507.18956 , archivePrefix =. 2507.18956 , primaryClass =

    doi

  92. [100]

    2024 , eprint=

    Chain of Agents: Large Language Models Collaborating on Long-Context Tasks , author=. 2024 , eprint=

    2024

  93. [101]

    arXiv e-prints , keywords =

    Language Models (Mostly) Do Not Consider Emotion Triggers When Predicting Emotion. arXiv e-prints , keywords =. doi:10.48550/arXiv.2311.09602 , archivePrefix =. 2311.09602 , primaryClass =

    doi arXiv

  94. [102]

    arXiv e-prints , keywords =

    Lost in the Middle: How Language Models Use Long Contexts. arXiv e-prints , keywords =. doi:10.48550/arXiv.2307.03172 , archivePrefix =. 2307.03172 , primaryClass =

    doi arXiv

  95. [103]

    2023 , eprint=

    LLaMA: Open and Efficient Foundation Language Models , author=. 2023 , eprint=

    2023

  96. [104]

    2024 IEEE International Conference on Big Data (BigData) , year=

    Evaluating Text Summarization Techniques and Factual Consistency with Language Models , author=. 2024 IEEE International Conference on Big Data (BigData) , year=

    2024

  97. [105]

    2025 , eprint=

    An Evaluation of Large Language Models on Text Summarization Tasks Using Prompt Engineering Techniques , author=. 2025 , eprint=

    2025

  98. [106]

    Journal of King Saud University Computer and Information Sciences , year =

    Chen, Xiaoyong and Chen, Zhiqiang and Cheng, Shi , title =. Journal of King Saud University Computer and Information Sciences , year =. doi:10.1007/s44443-025-00041-2 , url =

    doi

  99. [107]

    and Zhang, Jiawei , title =

    Zhang, Haopeng and Yu, Philip S. and Zhang, Jiawei , title =. ACM Comput. Surv. , month = jun, articleno =. 2025 , issue_date =. doi:10.1145/3731445 , abstract =

    2025 doi

  100. [108]

    A review of affective computing: From unimodal analysis to multimodal fusion , journal =

    Soujanya Poria and Erik Cambria and Rajiv Bajpai and Amir Hussain , keywords =. A review of affective computing: From unimodal analysis to multimodal fusion , journal =. 2017 , issn =. doi:https://doi.org/10.1016/j.inffus.2017.02.003 , url =

    2017 doi

  101. [109]

    2024 , eprint=

    Emotion-LLaMA: Multimodal Emotion Recognition and Reasoning with Instruction Tuning , author=. 2024 , eprint=

    2024

  102. [110]

    ICASSP , pages=

    IEMOCAP: Interactive emotional dyadic motion capture database , author=. ICASSP , pages=

  103. [111]

    arXiv e-prints , keywords =

    Spoken DialogSum: An Emotion-Rich Conversational Dataset for Spoken Dialogue Summarization. arXiv e-prints , keywords =. doi:10.48550/arXiv.2512.14687 , archivePrefix =. 2512.14687 , primaryClass =

    doi

  104. [112]

    2024 , eprint=

    The Llama 3 Herd of Models , author=. 2024 , eprint=

    2024

  105. [113]

    2020 , eprint=

    Fill in the BLANC: Human-free quality estimation of document summaries , author=. 2020 , eprint=

    2020

  106. [114]

    Soviet Physics Doklady , year = 1966, month = feb, volume =

    Binary Codes Capable of Correcting Deletions, Insertions and Reversals. Soviet Physics Doklady , year = 1966, month = feb, volume =

    1966

  107. [115]

    and Della Pietra, Vincent J

    Brown, Peter F. and Della Pietra, Vincent J. and deSouza, Peter V. and Lai, Jenifer C. and Mercer, Robert L. Class-Based n-gram Models of Natural Language. Computational Linguistics. 1992

    1992

  108. [116]

    1901 , publisher=

    Etude comparative de la distribution florale dans une portion des Alpes et du Jura , author=. 1901 , publisher=

    1901

  109. [117]

    1983 , url=

    Introduction to Modern Information Retrieval , author=. 1983 , url=

    1983

  110. [118]

    MeetingBank: A Benchmark Dataset for Meeting Summarization

    Yebowen Hu and Tim Ganter and Hanieh Deilamsalehy and Franck Dernoncourt and Hassan Foroosh and Fei Liu. MeetingBank: A Benchmark Dataset for Meeting Summarization. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL). 2023

    2023

  111. [119]

    and Lee, Sungbok and Narayanan, Shrikanth S

    Busso, Carlos and Bulut, Murtaza and Lee, Chi-Chun and Kazemzadeh, Abe and Mower, Emily and Kim, Samuel and Chang, Jeannette N. and Lee, Sungbok and Narayanan, Shrikanth S. , biburl =. IEMOCAP: interactive emotional dyadic motion capture database. , url =. Language Resources a...

  112. [120]

    MELD : A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations

    Poria, Soujanya and Hazarika, Devamanyu and Majumder, Navonil and Naik, Gautam and Cambria, Erik and Mihalcea, Rada. MELD : A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations. Proceedings of the 57th Annual Meeting of the Association for Computational Li...

    2019 doi

  113. [121]

    D ial S umm E val: Revisiting Summarization Evaluation for Dialogues

    Gao, Mingqi and Wan, Xiaojun. D ial S umm E val: Revisiting Summarization Evaluation for Dialogues. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022. doi:10.18653/v1/2022.naacl-main.418

    2022 doi

  114. [122]

    2026 , eprint=

    Emotion-LLaMAv2 and MMEVerse: A New Framework and Benchmark for Multimodal Emotion Understanding , author=. 2026 , eprint=

    2026

  115. [123]

    2025 , eprint=

    Fine-tuning for Better Few Shot Prompting: An Empirical Comparison for Short Answer Grading , author=. 2025 , eprint=

    2025

  116. [124]

    2021 , eprint=

    Unsupervised Topic Segmentation of Meetings with BERT Embeddings , author=. 2021 , eprint=

    2021

  117. [125]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Topic-oriented spoken dialogue summarization for customer service with saliency-aware topic modeling , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  118. [126]

    arXiv preprint arXiv:2511.03718 , year=

    Grounded Misunderstandings in Asymmetric Dialogue: A Perspectivist Annotation Scheme for MapTask , author=. arXiv preprint arXiv:2511.03718 , year=

  119. [127]

    Proceedings of the First Workshop on Evaluation and Comparison of NLP Systems , pages=

    Fill in the BLANC: Human-free quality estimation of document summaries , author=. Proceedings of the First Workshop on Evaluation and Comparison of NLP Systems , pages=

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.