Pith. sign in

REVIEW 4 major objections 3 minor 3 cited by

ChEmbed: Enhancing Chemical Literature Search Through Domain-Specific Text Embeddings

T0 review · 4 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read ChEmbed claims a nine-point nDCG@10 gain over general embeddings on chemical literature retrieval by fine-tuning on 1.7 million synthetic query-passage pairs.

desk verdict Plausible +9-point chemical retrieval gain, but the abstract alone can't rule out same-corpus leakage; the missing full text and absent split details make this a referee's paper, not a desk reject. read the letter →

arxiv 2508.01643 v1 pith:V7Z7EG6B submitted 2025-08-03 cs.IR cs.CL

Ali Shiraee Kasmaee , Mohammad Khodadad , Mehdi Astaraki , Mohammad Arshi Saloot , Nicholas Sherck , Hamidreza Mahyar , Soheila Samiee This is my paper · ORCID
classification cs.IRcs.CL
keywords chemicalliteratureretrievaltextembeddingsretrieval-augmentedgenerationdomain-specificfine-tuningtokenizeraugmentationsyntheticquerylong-contextChemRxivbenchmark
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

ChEmbed is a family of text embedding models built for chemical literature. The paper's goal is to show that a reasonably small, domain-adapted embedding model can outperform general-purpose embeddings on chemistry retrieval, which matters because RAG systems inherit the quality of whatever retriever sits underneath them. The paper claims that fine-tuning on roughly 1.7 million LLM-generated query-passage pairs from PubChem, Semantic Scholar, and ChemRxiv, together with 900 added chemical tokens and an 8192-token context length, raises nDCG@10 from 0.82 to 0.91 on a new ChemRxiv Retrieval benchmark. If correct, this gives chemistry RAG builders a direct swap-in improvement without changing the generation model.

What carries the argument

Three mechanisms carry the argument: (1) synthetic query generation by LLMs over chemistry-specific text, which creates the supervision signal for contrastive fine-tuning at roughly 1.7 million pairs; (2) tokenizer augmentation with 900 chemically specialized tokens placed in previously unused slots, which reduces fragmentation of chemical entities such as IUPAC names; (3) an 8192-token context length, which lets passages far longer than the 512- or 2048-token windows of many open-source embedding models be encoded in one vector. The effectiveness measure that ties them together is nDCG@10 on the paper's ChemRxiv Retrieval benchmark.

What would settle it

Check every benchmark query and passage against the training corpora using exact-title matching and near-duplicate n-gram hashing; if a non-negligible share of benchmark content appears in the fine-tuning data, the +9 pp nDCG@10 gain would not be evidence of domain adaptation. Alternatively, re-run the benchmark on ChemRxiv documents published after the training corpus cutoff and see whether the improvement survives.

Watch

Extended reading notes

Core claim

The central claim is that domain-specific fine-tuning closes most of the retrieval gap that general text embeddings leave on chemical terminology. Evaluated on the newly introduced ChemRxiv Retrieval benchmark, ChEmbed outperforms state-of-the-art general embedding models, lifting normalized discounted cumulative gain at rank 10 (nDCG@10) from 0.82 to 0.91, a nine-percentage-point gain. The model is trained on chemistry-specific text from PubChem, Semantic Scholar, and ChemRxiv, with LLMs used to synthesize roughly 1.7 million query-passage pairs, and its tokenizer is extended by 900 chemically specialized tokens so that IUPAC names and similar entities fragment less. ChEmbed also keeps an 8192-token context window, which the paper argues enables efficient retrieval of longer passages than typical open-source embeddings with 512 or 2048 tokens. The paper presents the model, the training recipe, and the benchmark as one reproducible package.

Load-bearing premise

The load-bearing premise is that the ChemRxiv Retrieval benchmark is genuinely separate from, and harder than, the ChemRxiv text used to fine-tune ChEmbed, so the reported 0.82-to-0.91 gain measures real retrieval ability rather than overlap between training data and test data.

Editorial extensions

If this is right

  • Chemistry RAG systems could replace their general embedding retriever with ChEmbed and expect the ranking gain to appear without retraining the generation model.
  • With an 8192-token context window, retrieval can index full abstracts and longer method sections, which should reduce information loss for queries answered from long passages.
  • The 900 added tokenizer tokens should make IUPAC names and reagent strings embed more atomically, which is testable on entity-heavy chemistry queries.
  • The synthetic query generation pipeline offers a recipe for building domain-specific retrieval training sets in other sciences where real user queries are scarce.

Reading between the lines

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

  • The decisive question the paper leaves implicit is train/test separation for the ChemRxiv benchmark; an external overlap audit would settle whether the nine-point gain transfers to genuinely new chemistry documents.
  • The same tokenizer-extension trick may translate to other naming-dense fields such as biology or materials science, but the paper only demonstrates it for chemistry.
  • Because the benchmark and the fine-tuning corpus both come from ChemRxiv, the reported gain could be narrower on query types outside the corpus's distribution, such as patents or experimental lab notes.
  • A per-query breakdown of nDCG@10 would show whether the gain is uniform or concentrated on long, terminology-heavy queries.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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. This manuscript introduces ChEmbed, a family of text-embedding models fine-tuned on chemistry-specific text drawn from PubChem, Semantic Scholar, and ChemRxiv. The abstract reports that training uses roughly 1.7 million LLM-generated query-passage pairs, a tokenizer augmented with 900 chemistry tokens, and an 8192-token context length. The central claim is that, on a newly introduced ChemRxiv Retrieval benchmark, ChEmbed outperforms state-of-the-art general embedding models, raising nDCG@10 from 0.82 to 0.91. The submitted full text, however, is an unrelated emotion-recognition paper (arXiv:2508.01644), so the manuscript contains no methods section, no experimental details, and no dataset-construction or evaluation protocol. The only substantive content is the abstract, and the evaluation claim is presented without error bars, significance tests, named baselines, or a statement of train/test separation.

Significance. If the claims were substantiated, a lightweight chemistry-specific embedding model with an extended tokenizer and long context would be a practically useful contribution to chemical literature retrieval. The stated result of a nine-point nDCG@10 gain over general embeddings is potentially valuable. The paper also advertises reproducibility, which is a positive feature. However, the current submission provides no evidence beyond the abstract: no model details, no baseline table, no benchmark statistics, and no guarantee that the training corpora and the evaluation benchmark are disjoint. As a result, the significance cannot currently be assessed, and the core claim is unverifiable.

major comments (4)
  1. [Abstract] The central evaluation claim is not supported by any stated train/test separation. The abstract says ChEmbed was fine-tuned on ChemRxiv text and evaluated on a newly introduced ChemRxiv Retrieval benchmark, but it does not state whether the benchmark passages or queries are disjoint from the training data, whether any deduplication was performed, or whether a publication-date cutoff was enforced. Because the same authors constructed both the training corpus and the benchmark from the same source, the reported +9 pp nDCG@10 gain could be inflated by overlap or by query templates that paraphrase training passages. The manuscript must explicitly document the split construction, including any deduplication and the provenance of the queries.
  2. [Full text / Methodology] The full text supplied for arXiv:2508.01643 is an unrelated paper on multimodal emotion recognition (DRKF, arXiv:2508.01644), not the ChEmbed paper. Consequently, the manuscript contains no description of the model architecture, fine-tuning procedure, synthetic-query generation, tokenizer extension, baseline tuning, evaluation protocol, or benchmark curation. Without these components, the abstract's performance claims cannot be checked by a reader or by a referee. This is a load-bearing omission that prevents any meaningful assessment of the paper's correctness.
  3. [Abstract (experimental reporting)] The reported result 'nDCG@10 from 0.82 to 0.91' is presented without any measure of variance, number of test queries, or significance test. It is therefore impossible to know whether the nine-point improvement is reliable or within noise. The manuscript should report standard deviations or confidence intervals across query sets, the size of the evaluation set, and the results of significance tests (e.g., paired bootstrap or permutation tests on nDCG@10).
  4. [Abstract (baselines)] The claim that ChEmbed 'outperforms state-of-the-art general embedding models' is not supported by a named baseline table. The abstract gives only a single aggregate number for the best baseline and does not identify which models were compared, how their hyperparameters were tuned, or what context lengths were used. A proper comparison should include per-model scores, model versions, and the retrieval setup (e.g., dense-only vs. hybrid, reranking, and indexing details).
minor comments (3)
  1. [Abstract (tokenizer)] The statement that the tokenizer adds '900 chemically specialized tokens to previously unused slots' is unclear: it should specify the base tokenizer, how the tokens were selected, and whether the effect of the tokenizer extension was isolated from the fine-tuning in an ablation.
  2. [Abstract (corpora)] The description of training data as 'chemistry-specific text from the PubChem, Semantic Scholar, and ChemRxiv corpora' is too coarse; the manuscript should state whether full texts, abstracts, or titles were used and how the 1.7 million query-passage pairs were filtered for quality and deduplicated.
  3. [General] The paper would benefit from a clear statement of the intended use cases and limitations, including failure modes for queries containing non-IUPAC nomenclature or non-text chemical structures, and a discussion of the computational cost of fine-tuning.

Circularity Check

0 steps flagged · score 0.0 of 10

No demonstrable circularity in the abstract; the benchmark-overlap concern is an empirical validity risk, not a shown reduction in the derivation chain.

full rationale

Based on the supplied abstract alone, ChEmbed's central claim is an empirical evaluation result, not a quantity derived by construction. The abstract says the model was 'fine-tuned on a dataset comprising chemistry-specific text from the PubChem, Semantic Scholar, and ChemRxiv corpora' and then 'Evaluated on our newly introduced ChemRxiv Retrieval benchmark,' but it does not state that the benchmark passages or LLM-generated queries are the same items used for fine-tuning. Sharing the ChemRxiv source name is not itself evidence that the training and test sets overlap; disjoint held-out evaluation from the same corpus is a standard and legitimate protocol in domain adaptation. No equation-level reduction, no fitted parameter renamed as a prediction, no load-bearing self-citation, no imported uniqueness theorem, and no ansatz smuggled in via citation can be exhibited from the text. The supplied full text is an unrelated paper and contains no ChEmbed assertions. The absence of an explicit train/test split is a missing detail and a possible leakage risk, but under the hard rules, a mere possibility of overlap is not the same as exhibiting a specific circular step. Therefore the honest finding is no significant circularity, and the score is 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

Because the full text is an unrelated paper, the ledger reflects only assumptions visible in the abstract. The main unverified assumptions are the validity of LLM-generated queries and the independence of the evaluation benchmark from the training corpus. No new physical or theoretical entities are introduced.

free parameters (2)
  • Fine-tuned model weights = not disclosed
    All reported retrieval performance depends on weights trained on 1.7 million synthetic query-passage pairs; the abstract gives no parameter count, training objective, or weight values.
  • Tokenizer vocabulary extension = 900 tokens
    The number of added chemistry tokens is a hand-chosen design choice; the abstract gives no criterion or ablation for this count.
assumptions (3)
  • domain assumption LLM-generated queries are a valid proxy for real user retrieval queries
    The abstract says queries were generated synthetically with large language models; retrieval quality on real chemistry searches depends on this assumption.
  • domain assumption ChemRxiv Retrieval benchmark is representative and disjoint from the training data
    Fine-tuning used ChemRxiv text and evaluation uses a ChemRxiv benchmark; unless the benchmark is held out and representative, the reported nDCG gain may reflect leakage.
  • domain assumption Adding 900 chemical tokens materially reduces fragmentation and improves retrieval
    The abstract asserts tokenizer augmentation helps without providing an ablation or evidence that fragmentation is the binding constraint.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ChEmbed: Enhancing Chemical Literature Search Through Domain-Specific Text Embeddings." pith.science (2026). https://pith.science/paper/V7Z7EG6B

@misc{pith2026250801643,
  author       = {Pith},
  title        = {Pith review of: ChEmbed: Enhancing Chemical Literature Search Through Domain-Specific Text Embeddings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V7Z7EG6B}},
  note         = {Machine review of arXiv:2508.01643}
}
read the original abstract

Retrieval-Augmented Generation (RAG) systems in chemistry heavily depend on accurate and relevant retrieval of chemical literature. However, general-purpose text embedding models frequently fail to adequately represent complex chemical terminologies, resulting in suboptimal retrieval quality. Specialized embedding models tailored to chemical literature retrieval have not yet been developed, leaving a substantial performance gap. To address this challenge, we introduce ChEmbed, a domain-adapted family of text embedding models fine-tuned on a dataset comprising chemistry-specific text from the PubChem, Semantic Scholar, and ChemRxiv corpora. To create effective training data, we employ large language models to synthetically generate queries, resulting in approximately 1.7 million high-quality query-passage pairs. Additionally, we augment the tokenizer by adding 900 chemically specialized tokens to previously unused slots, which significantly reduces the fragmentation of chemical entities, such as IUPAC names. ChEmbed also maintains a 8192-token context length, enabling the efficient retrieval of longer passages compared to many other open-source embedding models, which typically have a context length of 512 or 2048 tokens. Evaluated on our newly introduced ChemRxiv Retrieval benchmark, ChEmbed outperforms state-of-the-art general embedding models, raising nDCG@10 from 0.82 to 0.91 (+9 pp). ChEmbed represents a practical, lightweight, and reproducible embedding solution that effectively improves retrieval for chemical literature search.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. When Iterative RAG Beats Ideal Evidence: A Diagnostic Study in Scientific Multi-hop Question Answering

    cs.CL 2026-01 conditional novelty 7.0 of 10

    On ChemKGMultiHopQA, iterative retrieval-reasoning outperformed oracle gold-context static RAG for all 11 LLMs tested, with gains up to 25.6 percentage points.

  2. Bi-semantic Chemical Embedder for Joint Representation Learning of SMILES and Natural Language

    cs.LG 2026-08 conditional novelty 6.0 of 10

    A ModernBERT-based encoder trained with masked language modeling on SMILES-annotated scientific documents plus a contrastive stage yields embeddings that are competitive on both molecular property prediction and scien...

  3. GAPrune: Gradient-Alignment Pruning for Domain-Aware Embeddings

    cs.CL 2025-09 conditional novelty 6.0 of 10

    GAPrune combines Fisher information with cross-domain gradient alignment to prune embedding models, staying within 2.5% of dense at 50% sparsity and improving after retraining.

Reference graph

Works this paper leans on

31 extracted references · 19 canonical work pages · cited by 3 Pith papers

  1. [29]

    Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Xiangzhan Yu, and Furu Wei. 2022. WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing. ���� ������� �� �������� ������ �� ��...

    arXiv 2022

  2. [30]

    Ronghao Lin and Haifeng Hu. 2024. Adapt and explore: Multimodal mixup for representation learning. ����������� ������105 (2024), 102216. doi:10.1016/j. inffus.2023.102216

    arXiv 2024

  3. [32]

    Zhirong Wu, Yuanjun Xiong, Stella X Yu, and Dahua Lin. 2018. Unsupervised Feature Learning via Non-Parametric Instance Discrimination. In ����������� �� ��� ���� ���������� �� �������� ������ ��� ������� ����������� ������. IEEE, Salt Lake City, UT, USA, 3733–3742

    work page 2018

  4. [33]

    Yu-An Chung, Wei-Hung Weng, Schrasing Tong, and James Glass. 2018. Un- supervised Cross-Modal Alignment of Speech and Text Embedding Spaces. In �������� �� ������ ����������� ���������� �������, Vol. 31. Curran Associates, Inc., Red Hook, NY, USA, 7354–7365. https://proceedings.neurips.cc/paper_files/ paper/2018/file/f3ce96dfe0061d0e6e105b0b70e5aafb-Paper.pdf

    work page 2018

  5. [34]

    Mohamed Ishmael Belghazi, Aristide Baratin, Sai Rajeshwar, Sherjil Ozair, Yoshua Bengio, Aaron Courville, and Devon Hjelm. 2018. Mutual Informa- tion Neural Estimation. In ����������� �� ��� ���� ������������� ���������� �� ������� �������� ������������ �� ������� �������� ��������� ���� ���, Jennifer Dy and Andreas Krause (Eds.). PMLR, Stockholm, Sweden,...

    work page 2018

  6. [35]

    Pengyu Cheng, Weituo Hao, Shuyang Dai, Jiachang Liu, Zhe Gan, and Lawrence Carin. 2020. CLUB: A Contrastive Log-Ratio Upper Bound of Mutual Information. In ����������� �� ��� ���� ������������� ���������� �� ������� �������� ���������. JMLR.org, Virtual Event, Article 166, 10 pages

    work page 2020

  7. [36]

    Tadas Baltrušaitis, Chaitanya Ahuja, and Louis-Philippe Morency. 2019. Multi- modal Machine Learning: A Survey and Taxonomy. ���� ������������ �� ������� �������� ��� ������� ������������41, 2 (2019), 423–443. doi:10.1109/TPAMI.2018. 2798607

    doi:10.1109/tpami.2018 2019

  8. [37]

    Devamanyu Hazarika, Soujanya Poria, Rada Mihalcea, Erik Cambria, and Roger Zimmermann. 2018. ICON: Interactive Conversational Memory Network for Multimodal Emotion Detection. In����������� �� ��� ���� ���������� �� ��������� ������� �� ������� �������� ����������, Ellen Riloff, David Chiang, Julia Hock- enmaier, and Jun’ichi Tsujii (Eds.). Association for...

    doi:10.18653/v1/d18-1280 2018

Show all 31 references
  1. [38]

    Efthymios Georgiou, Charilaos Papaioannou, and Alexandros Potamianos. 2019. Deep Hierarchical Fusion with Application in Sentiment Analysis. In ����������� �� ����������� ����. ISCA, Graz, Austria, 3302–3306. https://api.semanticscholar. org/CorpusID:202736442

    2019

  2. [39]

    Qiuju Zhang, Hongtao Zhang, Keming Zhou, and Le Zhang. 2023. Developing a Physiological Signal-Based, Mean Threshold and Decision-Level Fusion Algo- rithm (PMD) for Emotion Recognition. �������� ������� ��� ����������28, 4 (2023), 673–685. doi:10.26599/TST.2022.9010038

    2023

  3. [40]

    Seunghyun Yoon, Seokhyun Byun, Subhadeep Dey, and Kyomin Jung. 2019. Speech Emotion Recognition Using Multi-hop Attention Mechanism. In ������ ���� � ���� ���� ������������� ���������� �� ���������� ������ ��� ������ ���������� ��������. IEEE, Brighton, United Kingdom, 2822–28...

    2019 doi

  4. [41]

    Pengfei Liu, Kun Li, and Helen Meng. 2020. Group Gated Fusion on Attention- Based Bidirectional Alignment for Multimodal Emotion Recognition. In ���� �������� �� ����������� ����. ISCA, Shanghai, China, 379–383. doi:10.21437/ Interspeech.2020-2067

    2020

  5. [42]

    Lin Feng, Lu-Yao Liu, Sheng-Lan Liu, Jian Zhou, Han-Qing Yang, and Jie Yang

  6. [43]

    Mustaqeem Khan, Wail Gueaieb, Abdulmotaleb El Saddik, and Soonil Kwon. 2024. MSER: Multimodal speech emotion recognition using cross-attention with deep fusion. ������ ������� ���� ������������245 (2024), 122946. doi:10.1016/j.eswa. 2023.122946

    2024

  7. [44]

    Xiaoheng Zhang and Yang Li. 2023. A Dual Attention-based Modality- Collaborative Fusion Network for Emotion Recognition. In ����������� �� ������ ������ ����. ISCA, Dublin, Ireland, 1468–1472. doi:10.21437/Interspeech.2023-523

    2023 doi

  8. [45]

    Mixiao Hou, Zheng Zhang, Chang Liu, and Guangming Lu. 2023. Semantic Alignment Network for Multi-Modal Emotion Recognition. ���� ������������ �� �������� ��� ������� ��� ����� ����������33, 9 (2023), 5318–5329. doi:10.1109/ TCSVT.2023.3247822

    2023

  9. [46]

    Jennifer Santoso, Kenkichi Ishizuka, and Taiichi Hashimoto. 2024. Large Lan- guage Model-Based Emotional Speech Annotation Using Context and Acoustic Feature for Speech Emotion Recognition. In������ ���� � ���� ���� ������������� ���������� �� ���������� ������ ��� ������ ����...

    2024

  10. [48]

    Zihan Zhao, Yu Wang, and Yanfeng Wang. 2023. Knowledge-aware Bayesian Co-attention for Multimodal Emotion Recognition. In ������ ���� � ���� ���� ������������� ���������� �� ���������� ������ ��� ������ ���������� ��������. IEEE, Rhodes Island, Greece, 1–5

    2023

  11. [49]

    Yufan Yi, Yan Tian, Cong He, Yajing Fan, Xinli Hu, and Yiping Xu. 2023. DBT: multimodal emotion recognition based on dual-branch transformer. ��� ������� �� ��������������79, 8 (2023), 8611–8633. doi:10.1007/s11227-022-05001-5

    2023 doi

  12. [50]

    Carlos Busso, Murtaza Bulut, Chi-Chun Lee, Abe Kazemzadeh, Emily Mower, Samuel Kim, Jeannette N Chang, Sungbok Lee, and Shrikanth S Narayanan. 2008. IEMOCAP: Interactive emotional dyadic motion capture database. �������� ��������� ��� ����������42 (2008), 335–359. doi:10.1007/...

    2008 doi

  13. [51]

    Soujanya Poria, Devamanyu Hazarika, Navonil Majumder, Gautam Naik, Erik Cambria, and Rada Mihalcea. 2019. MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations. In ����������� �� ��� ���� ������ ������� �� ��� ����������� ��� ������������� �����������...

    2019 doi

  14. [52]

    Jinming Zhao, Tenggan Zhang, Jingwen Hu, Yuchen Liu, Qin Jin, Xinchao Wang, and Haizhou Li. 2022. M3ED: Multi-modal Multi-scene Multi-label Emotional Dialogue Database. In����������� �� ��� ���� ������ ������� �� ��� ����������� ��� ������������� ����������� ������� �� ���� ��...

    2022 doi

  15. [53]

    Jintao Wen, Dazhi Jiang, Geng Tu, Cheng Liu, and Erik Cambria. 2023. Dynamic interactive multiview memory network for emotion recognition in conversation. ����������� ������91 (2023), 123–133. doi:10.1016/j.inffus.2022.10.009

    2023 doi

  16. [55]

    Lili Guo, Yikang Song, and Shifei Ding. 2024. Speaker-aware cognitive network with cross-modal attention for multimodal emotion recognition in conversation. ��������������� �������296 (2024), 111969. doi:10.1016/j.knosys.2024.111969

    2024

  17. [57]

    Baijun Xie, Mariia Sidulova, and Chung Hyuk Park. 2021. Robust Multimodal Emotion Recognition from Conversation with Transformer-Based Crossmodality Fusion. �������21, 14, Article 4913 (2021), 16 pages. doi:10.3390/s21144913

    2021 doi

  18. [58]

    Zheng Lian, Bin Liu, and Jianhua Tao. 2023. SMIN: Semi-Supervised Multi-Modal Interaction Network for Conversational Emotion Recognition. ���� ������������ �� �������� ���������14, 3 (2023), 2415–2429. doi:10.1109/TAFFC.2022.3141237

    2023

  19. [59]

    Changzeng Fu, Fengkui Qian, Kaifeng Su, Yikai Su, Ze Wang, Jiaqi Shi, Zhigang Liu, Chaoran Liu, and Carlos Toshinori Ishi. 2025. HiMul-LGG: A hierarchical decision fusion-based local–global graph neural network for multimodal emotion recognition in conversation. ������ �������...

    2025

  20. [60]

    Zixuan Peng, Yu Lu, Shengfeng Pan, and Yunfeng Liu. 2021. Efficient Speech Emotion Recognition Using Multi-Scale CNN and Attention. In������ ���� � ���� ���� ������������� ���������� �� ���������� ������ ��� ������ ���������� ��������. IEEE, Toronto, ON, Canada, 3020–3024. doi...

    2021

  21. [61]

    Soumya Dutta and Sriram Ganapathy. 2022. Multimodal Transformer with Learnable Frontend and Self Attention for Emotion Recognition. In ������ ���� � ���� ���� ������������� ���������� �� ���������� ������ ��� ������ ���������� ��������. IEEE, Singapore, 6917–6921. doi:10.1109/...

    2022

  22. [62]

    Cheng Peng, Ke Chen, Lidan Shou, and Gang Chen. 2024. CARAT: Contrastive Feature Reconstruction and Aggregation for Multi-Modal Multi-Label Emotion Recognition. In ����������� �� ��� ������������� ���� ���������� �� ��������� ������������ ��� ������������ ���������� �� �������...

    2024 doi

  23. [2023]

    ���������� ����� �����82, 19 (March 2023), 28917–28935

    Multimodal speech emotion recognition based on multi-scale MFCCs and multi-view attention mechanism. ���������� ����� �����82, 19 (March 2023), 28917–28935. doi:10.1007/s11042-023-14600-0

    2023 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.