Pith. sign in

REVIEW 4 major objections 5 minor 36 references

Towards an Automated Multimodal Approach for Video Summarization: Building a Bridge Between Text, Audio and Facial Cue-Based Summarization

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that scoring transcript sentences with prosodic and facial emphasis cues, not just word frequency, lifts timestamped video summaries from 0.4769 to 0.7929 ROUGE-1 and improves segment-selection F1 by about 23 percent on…

desk verdict The bonus-word idea is worth a look, but the evaluation is confounded by summary length and a self-generated LLM reference, so the headline numbers don't carry. read the letter →

arxiv 2506.23714 v1 pith:Y43AAPGI submitted 2025-06-30 cs.CV cs.CL

Md Moinul Islam , Sofoklis Kakouros , Janne Heikkilä , Mourad Oussalah This is my paper · ORCID
classification cs.CVcs.CL
keywords multimodalvideosummarizationcomputervisionnaturallanguageprocessingaudioandspeechmultimediacontentcreationbonuswordsprosodicfeaturespseudo-groundtruth
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

This paper tries to establish that a behavior-aware, multimodal video summarization pipeline—fusing text keywords with prosodic and facial-emphasis cues—produces better timestamped summaries than text-only extractive methods. The motivation is practical: interviews, lectures, and meetings carry meaning in tone and expression, not just words, and conventional summarizers ignore those signals. The authors evaluate on 1,500 short interview clips from the ChaLearn First Impressions dataset, using LLM-generated extractive pseudo-ground-truth summaries as references, and report large gains over the Edmundson baseline across text and video metrics. A sympathetic reader would take the paper's contribution to be the 'bonus word' mechanism and the demonstration that multimodal emphasis aligns with summarization importance.

What carries the argument

The load-bearing mechanism is the bonus-word scoring pipeline. A unified timeline, built with forced alignment, fuses textual cues (TF-IDF, sentiment analysis, named entities), prosodic cues (pitch, RMS loudness, Hammarberg index), and visual cues (nose-landmark displacement, facial emotion transitions); each transcript sentence is weighted by the frequency of bonus words, selected with an adaptive threshold $\theta = \mu + \lambda\sigma$, and diversified through a TF-IDF cosine-similarity penalty. The selected sentences are mapped to frame ranges and compiled into a subtitled summary video, while a separate GPT-4.5 extractive prompt generates pseudo-ground-truth summaries from the same timestamped transcripts for evaluation.

What would settle it

Take a random sample of the same interview videos, have human annotators mark the most important sentences and segments, and recompute ROUGE, BERTScore, and segment F1 against those human references. If the multimodal method no longer beats Edmundson, or if its advantage shrinks to statistical noise, the paper's central claim is falsified. A direct check of the verbosity confound would be to equalize summary length between the two methods before comparing, since the proposed summaries are nearly twice as long (length ratio 0.5734 vs 0.3125), which can mechanically raise recall-style metrics such as ROUGE-1.

Watch

Extended reading notes

Core claim

The central claim is that a word's summarization importance is not purely lexical: when a transcript word coincides in time with a prosodic spike (pitch, loudness, tonality) or a visual cue (head movement, facial emotion transition), it becomes a 'bonus word' and should be weighted more heavily in sentence scoring. Combining this multimodal weighting with an adaptive threshold and a diversity penalty produces text summaries and timestamped video segments that, on the ChaLearn interview-clip subset, beat Edmundson's classical keyword-position-cue method on every reported metric: ROUGE-1 0.7929 vs 0.4769, BERTScore 0.9536 vs 0.9152, BLEU 0.6411 vs 0.2259, and video segment F1 0.6995 vs 0.5709.

Load-bearing premise

The load-bearing premise is that the GPT-generated pseudo-ground-truth summaries are a valid stand-in for human judgments of which sentences matter; if those references are biased, the reported improvements do not show better summarization.

Editorial extensions

If this is right

  • If the framework generalizes to other conversational video, timestamped summaries will preserve emotional and prosodic emphasis rather than only lexical content.
  • The ablation indicates visual cues are the strongest single modality and removing them hurts performance most, so systems without face and pose analysis will underperform on expressive interview-style content.
  • The bonus-word mechanism offers a lightweight way to inject multimodal salience into classic extractive scoring without training a deep model.
  • The pseudo-ground-truth evaluation strategy provides a scalable path for assessing video summarization when human annotations are unavailable, subject to the paper's own caveats about validation.

Reading between the lines

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

  • Beyond the paper: the reported ROUGE gains could be partly driven by verbosity, since the proposed summaries are nearly twice as long as Edmundson's (length ratio 0.5734 vs 0.3125), so an exact-length-matched comparison would separate content selection from summary length.
  • Beyond the paper: the bonus-word mechanism should transfer to meetings, lectures, and vlogs where emphasis signals align with spoken keywords, but longer-form videos may stress the timestamp-alignment precision and threshold choices.
  • Beyond the paper: a natural next test is human evaluation of summary usefulness and fluency, because the pseudo-ground-truth references capture only extractive sentence selection, not whether viewers actually find the summaries more informative or expressive.
  • Beyond the paper: the fixed threshold factor $\lambda=0.3$ and diversity factor $\delta=0.2$ deserve sensitivity analysis, since the claimed improvements could depend on these specific parameter values.
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 / 5 minor

Summary. This paper proposes a multimodal, extractive video summarization pipeline for short interview videos. It combines textual cues (TF-IDF, sentiment, named entities), audio prosodic cues (pitch, loudness, voice quality), and visual cues (head movement, facial emotion transitions) into 'bonus words' that are weighted when scoring sentences. An adaptive threshold (θ = μ + λσ) and a cosine-similarity diversity penalty select sentences, which are then mapped to timestamped video segments. The system is evaluated on 1,500 clips from the ChaLearn First Impressions dataset against pseudo-ground-truth (pGT) summaries generated by GPT-4.5 with an extractive prompt. The authors report large gains over the Edmundson baseline in ROUGE, BLEU, BERTScore, and video-selection F1, and they include an ablation study. The paper acknowledges limitations of pGT evaluation and reports a small qualitative validation and a consistency check.

Significance. If the central claim were established, the paper would offer a cheap, modular way to inject behavioral cues into extractive summarization, and the notion of 'bonus words' is an interesting operationalization of cross-modal emphasis. The manuscript describes its pipeline clearly and reuses publicly available components (Whisper, MFA, MediaPipe, DeepFace, openSMILE), which supports reproducibility. It also explicitly acknowledges the limitations of pseudo-ground-truth evaluation and attempts a consistency check. However, the significance is currently limited because the evaluation does not isolate the effect of multimodal cues from summary length, and the reference standard is self-generated rather than independently human-validated.

major comments (4)
  1. [§4.2, Tables 1–2] The central comparative claim is confounded by summary length. The proposed method has length ratio 0.5734 versus 0.3125 for Edmundson, i.e., approximately 83% longer. ROUGE-1/2/L, BLEU, and BERTScore are all sensitive to candidate length: a longer candidate has more opportunities to match pGT tokens and phrases. No length-matched baseline, compression-adjusted metric, or precision-recall curve is reported. The video F1 comparison is similarly affected because matches defined by IoU > 0.5 reward longer selected intervals. Therefore the abstract and Section 4.2 claim that the framework 'significantly outperforms' Edmundson is not supported by the reported numbers.
  2. [§4.1] All text and video metrics are computed against pseudo-ground-truth summaries produced by GPT-4.5 with an extractive prompt. Because the proposed method is also a sentence-extractive selector, the evaluation essentially measures agreement with the LLM's selection rather than summarization quality. The two-part validation (50 manually inspected summaries; average Jaccard similarity of 0.78 between prompt variants) is helpful but does not replace human relevance judgments or an external benchmark. Without such a reference, the claimed improvement is not established; I am not claiming the bonus-word construction is circular, but the benchmark is self-generated and may be biased.
  3. [§4.3, Table 3] The ablation results contradict the narrative. The multimodal system has BLEU 0.6411, while Visual-only achieves 0.7234 and w/o Audio achieves 0.7456; the statement that 'all single-modality variants clearly underperform compared to multimodal approaches' is false for BLEU. The absence of standard deviations and significance tests means that other differences (e.g., ROUGE-1 0.6743 vs. 0.7929) could also be within noise. Please report per-video or per-fold variance, significance tests, and length ratios for all variants.
  4. [§3, Algorithm 1, Eq. (2)] Several free parameters are hand-set without validation: λ = 0.3 in Eq. (2), the diversity factor δ = 0.2, the head-movement displacement threshold, and the audio Z-score threshold. These directly control how many sentences are selected and thus the summary length. Without a validation split, sensitivity analysis, or parameter sweeps, the reported gains could be artifacts of threshold choices. In addition, λ is used both for the head-movement threshold and for the adaptive threshold factor, which is confusing.
minor comments (5)
  1. [Figure 1 caption] The caption refers to 'ChaLearn'17', but the cited dataset is the ChaLearn LAP 2016 First Impressions challenge; please make the year consistent.
  2. [Section 3, 'Bonus words'] The definition of bonus words is informal: it is unclear whether a sentence weight is the number of bonus-word occurrences or the number of distinct bonus words; Algorithm 1's Count(w, Words) suggests the former, which can double-count repeated terms.
  3. [§4.1] The consistency check reports an average Jaccard similarity of 0.78 but does not state the number of videos sampled or the exact prompt variations used; please provide this detail.
  4. [Tables 1 and 3] Table 3 does not report length ratios for the ablation variants, making it impossible to interpret BLEU/ROUGE differences across variants, especially in light of the length confounding raised in the major comments.
  5. [§4.2] There is a typo 'psuedo-ground truth' in the video-based evaluation paragraph; it should read 'pseudo-ground truth'.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular reduction; the multimodal derivation is independent of the pGT reference, though the self-generated benchmark and length mismatch are validity concerns.

full rationale

No load-bearing circular step is present. The proposed summary is produced by Algorithm 1 from multimodal cues: bonus words are defined by temporal coincidence of textual terms with audio/visual salience ('A bonus word is selected when a textual term coincides temporally with salient cues in at least one additional modality'), and sentence weights are the count of such words. The pGT reference is generated by GPT-4.5 from transcripts only using an extractive prompt, so the reference is not defined in terms of the proposed bonus-word scoring, and no parameter is fitted to the pGT; lambda and delta are fixed hand-set constants. Consequently, the ROUGE/BLEU/F1 gains are not forced by construction. The self-generated pGT is a validity limitation, and the paper candidly says it 'cannot fully substitute human-annotated ground truth' (Section 4.1); likewise, the length-ratio difference (0.5734 vs 0.3125) is a plausible confound in the metric comparison, but a confound is not a definitional equivalence. The only self-citations ([8], and [4] via a co-author) support side claims about abstractive summarization and multimodal limits, not the central derivation. Overall circularity burden is 2/10: minor non-load-bearing self-citation, no circular reduction.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The framework depends on several hand-set thresholds, an unvalidated assumption that LLM-selected sentences are ground truth, and unstated assumptions that prosodic and facial signals mark importance. The bonus-word construct is the only invented concept and it lacks independent validation.

free parameters (5)
  • Adaptive selection threshold lambda = 0.3
    Controls sentence selection in Eq. 2; hand-set; no validation split described; directly changes summary length and all ROUGE/F1 numbers.
  • Diversity factor delta = 0.2
    Diversity check in Algorithm 1 line 25; hand-set; affects which sentences are excluded due to cosine similarity.
  • Head movement displacement threshold = not specified
    Visual cue detection uses a predefined threshold on nose landmark displacement; without a stated value the visual emphasis signal is not reproducible.
  • Audio cue Z-score threshold = not specified
    Audio cues are flagged when normalized pitch/loudness values exceed a predefined threshold; the value is not reported.
  • Data subset size = 1500
    The authors selected 15% of ChaLearn without describing the selection rule or seed; results may depend on this subset.
assumptions (5)
  • domain assumption GPT-4.5 extractive sentence selection is a valid proxy for human summary ground truth.
    Section 4.1 uses LLM-generated pGT as the evaluation reference while acknowledging limitations; all headline metrics depend on this.
  • domain assumption Visual cues from MediaPipe nose displacement and DeepFace emotion transitions indicate meaningful emphasis.
    Section 3 Visual Cues Detection; no validation that these events align with human-judged importance.
  • domain assumption Prosodic features (pitch, loudness, HammarbergIndex) signal communicative importance.
    Section 3 Audio Cues Detection; assumed without evidence.
  • domain assumption MFA word-level timestamps are accurate enough for sentence-to-video alignment.
    Section 3 Audio Processing; alignment underpins bonus-word detection and video segment extraction.
  • domain assumption ChaLearn First Impressions clips, originally for personality assessment, are a suitable domain for video summarization evaluation.
    Section 3 Data Sources; repurposing is asserted rather than justified.
invented entities (1)
  • Bonus words
    purpose: Terms emphasized across text, audio, or visual channels, used to weight sentences for extractive summarization.
    Introduced in Section 3 Cross-Modal Cues Alignment; no external validation that bonus-word selection captures human-perceived emphasis, only qualitative examples and aggregate metrics against a machine-generated reference.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards an Automated Multimodal Approach for Video Summarization: Building a Bridge Between Text, Audio and Facial Cue-Based Summarization." pith.science (2026). https://pith.science/paper/Y43AAPGI

@misc{pith2026250623714,
  author       = {Pith},
  title        = {Pith review of: Towards an Automated Multimodal Approach for Video Summarization: Building a Bridge Between Text, Audio and Facial Cue-Based Summarization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y43AAPGI}},
  note         = {Machine review of arXiv:2506.23714}
}
read the original abstract

The increasing volume of video content in educational, professional, and social domains necessitates effective summarization techniques that go beyond traditional unimodal approaches. This paper proposes a behaviour-aware multimodal video summarization framework that integrates textual, audio, and visual cues to generate timestamp-aligned summaries. By extracting prosodic features, textual cues and visual indicators, the framework identifies semantically and emotionally important moments. A key contribution is the identification of bonus words, which are terms emphasized across multiple modalities and used to improve the semantic relevance and expressive clarity of the summaries. The approach is evaluated against pseudo-ground truth (pGT) summaries generated using LLM-based extractive method. Experimental results demonstrate significant improvements over traditional extractive method, such as the Edmundson method, in both text and video-based evaluation metrics. Text-based metrics show ROUGE-1 increasing from 0.4769 to 0.7929 and BERTScore from 0.9152 to 0.9536, while in video-based evaluation, our proposed framework improves F1-Score by almost 23%. The findings underscore the potential of multimodal integration in producing comprehensive and behaviourally informed video summaries.

Figures

Figures reproduced from arXiv: 2506.23714 by the authors.

Figure 1
Figure 1. Multimodal Video Summarization Framework with Pseudo-Ground Truth Generation. The diagram illustrates a comprehensive pipeline for video summarization using the ChaLearn’17 dataset. Starting with input video, the process branches into parallel extraction paths: audio processing, frame extraction and speech-to-text conversion. The framework integrates audio, visual and textual cues extraction to generate a text summa… view at source ↗
Figure 2
Figure 2. Example of head movement deviations over time, visualized as a function of video frames. The blue line represents frame-to-frame head movement deviations, the green dashed line shows the moving average of these deviations and the red dotted line indicates the dynamic threshold calculated as the moving average plus 1.5 times the standard deviation. Peaks above the threshold suggest significant head movement events, w… view at source ↗
Figure 3
Figure 3. Example of pitch (𝐹0) variations over time. The blue line shows the interpolated pitch contour, and red dots mark significant changes based on a threshold, highlighting expressive moments relevant to summarization. Cross-Modal Cues Alignment. Cross-modal alignment integrates visual, audio, and textual cues into a unified timeline for our video summarization framework. Visual cues, such as pose and emotion, are align… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 31 canonical work pages

  1. [1]

    Apostolidis, E

    E. Apostolidis, E. Adamantidou, A. I. Metsai, V. Mezaris, I. Patras, Video summarization using deep neural networks: A survey, Proceedings of the IEEE 109 (2021) 1838–1863

    work page 2021

  2. [2]

    W.-S. Chu, Y. Song, A. Jaimes, Video co-summarization: Video summarization by visual co- occurrence, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 3584–3592

    work page 2015

  3. [3]

    Tiwari, C

    V. Tiwari, C. Bhatnagar, A survey of recent work on video summarization: approaches and techniques, Multimedia Tools and Applications 80 (2021) 27187–27221

    work page 2021

  4. [4]

    Otani, Y

    M. Otani, Y. Nakashima, E. Rahtu, J. Heikkilä, N. Yokoya, Video summarization using deep semantic features, in: Computer Vision–ACCV 2016: 13th Asian Conference on Computer Vision, Taipei, Taiwan, November 20-24, 2016, Revised Selected Papers, Part V 13, Springer, 2017, pp. 361–377

    work page 2016

  5. [5]

    Rochan, L

    M. Rochan, L. Ye, Y. Wang, Video summarization using fully convolutional sequence networks, in: Proceedings of the European conference on computer vision (ECCV), 2018, pp. 347–363

    work page 2018

  6. [6]

    A. P. Wibawa, F. Kurniawan, et al., A survey of text summarization: Techniques, evaluation and challenges, Natural Language Processing Journal 7 (2024) 100070

    work page 2024

  7. [7]

    H. P. Edmundson, New methods in automatic extracting, Journal of the ACM (JACM) 16 (1969) 264–285

    work page 1969

  8. [8]

    M. M. Islam, U. Muhammad, M. Oussalah, Evaluating text summarization techniques and factual consistency with language models, in: 2024 IEEE International Conference on Big Data (BigData), 2024, pp. 116–122. doi:10.1109/BigData62323.2024.10826032

    arXiv 2024

Show all 36 references
  1. [9]

    A. See, P. J. Liu, C. D. Manning, Get to the point: Summarization with pointer-generator networks, arXiv preprint arXiv:1704.04368 (2017)

    2017 arXiv

  2. [10]

    K. Zhou, Y. Qiao, T. Xiang, Deep reinforcement learning for unsupervised video summarization with diversity-representativeness reward, in: Proceedings of the AAAI conference on artificial intelligence, volume 32, 2018

    2018

  3. [11]

    T. Liu, Q. Meng, J.-J. Huang, A. Vlontzos, D. Rueckert, B. Kainz, Video summarization through reinforcement learning with a 3d spatio-temporal u-net, IEEE transactions on image processing 31 (2022) 1573–1586

    2022

  4. [12]

    Z. Li, L. Yang, Weakly supervised deep reinforcement learning for video summarization with se- mantically meaningful reward, in: Proceedings of the IEEE/CVF winter conference on applications of computer vision, 2021, pp. 3239–3247

    2021

  5. [13]

    G. Wang, X. Wu, J. Yan, Progressive reinforcement learning for video summarization, Information Sciences 655 (2024) 119888

    2024

  6. [14]

    Zhang, W.-L

    K. Zhang, W.-L. Chao, F. Sha, K. Grauman, Video summarization with long short-term memory, in: Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part VII 14, Springer, 2016, pp. 766–782

    2016

  7. [15]

    Saini, K

    P. Saini, K. Kumar, S. Kashid, A. Saini, A. Negi, Video summarization using deep learning techniques: a detailed analysis and investigation, Artificial Intelligence Review 56 (2023) 12347–12385

    2023

  8. [16]

    Evangelopoulos, A

    G. Evangelopoulos, A. Zlatintsi, A. Potamianos, P. Maragos, K. Rapantzikos, G. Skoumas, Y. Avrithis, Multimodal saliency and fusion for movie summarization based on aural, visual, and textual attention, IEEE Transactions on Multimedia 15 (2013) 1553–1568

    2013

  9. [17]

    J. Park, K. Kwoun, C. Lee, H. Lim, Multimodal frame-scoring transformer for video summarization, arXiv preprint arXiv:2207.01814 (2022)

    2022 arXiv

  10. [18]

    B. Zhao, M. Gong, X. Li, Audiovisual video summarization, IEEE Transactions on Neural Networks and Learning Systems 34 (2021) 5181–5188

    2021

  11. [19]

    B. Zhao, M. Gong, X. Li, Hierarchical multimodal transformer to summarize videos, Neurocom- puting 468 (2022) 360–369

    2022

  12. [20]

    Psallidas, P

    T. Psallidas, P. Koromilas, T. Giannakopoulos, E. Spyrou, Multimodal summarization of user- generated videos, Applied Sciences 11 (2021) 5260

    2021

  13. [21]

    Y. Zhu, W. Zhao, R. Hua, X. Wu, Topic-aware video summarization using multimodal transformer, Pattern Recognition 140 (2023) 109578

    2023

  14. [22]

    J. Xie, X. Chen, S. Zhao, S.-P. Lu, Video summarization via knowledge-aware multimodal deep networks, Knowledge-Based Systems 293 (2024) 111670

    2024

  15. [23]

    J. Xie, X. Chen, S.-P. Lu, Y. Yang, A knowledge augmented and multimodal-based framework for video summarization, in: Proceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 740–749

    2022

  16. [24]

    Ponce-López, B

    V. Ponce-López, B. Chen, M. Oliu, C. Corneanu, A. Clapés, I. Guyon, X. Baró, H. J. Escalante, S. Escalera, Chalearn lap 2016: First round challenge on first impressions-dataset and results, in: Computer Vision–ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 1...

    2016

  17. [25]

    Radford, J

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, I. Sutskever, Robust speech recognition via large-scale weak supervision, 2022. URL: https://arxiv.org/abs/2212.04356. arXiv:2212.04356

    2022 arXiv

  18. [26]

    Honnibal, I

    M. Honnibal, I. Montani, S. Van Landeghem, A. Boyd, et al., spacy: Industrial-strength natural language processing in python (2020)

    2020

  19. [27]

    McAuliffe, M

    M. McAuliffe, M. Socolof, S. Mihuc, M. Wagner, M. Sonderegger, Montreal forced aligner: Trainable text-speech alignment using kaldi., in: Interspeech, volume 2017, 2017, pp. 498–502

    2017

  20. [28]

    Bradski, The opencv library., Dr

    G. Bradski, The opencv library., Dr. Dobb’s Journal: Software Tools for the Professional Programmer 25 (2000) 120–123

    2000

  21. [29]

    Lugaresi, J

    C. Lugaresi, J. Tang, H. Nash, C. McClanahan, E. Uboweja, M. Hays, F. Zhang, C.-L. Chang, M. G. Yong, J. Lee, W.-T. Chang, W. Hua, M. Georg, M. Grundmann, Mediapipe: A framework for building perception pipelines, 2019. URL: https://arxiv.org/abs/1906.08172. arXiv:1906.08172

    2019 arXiv

  22. [30]

    Serengil, A

    S. Serengil, A. Ozpinar, A benchmark of facial recognition pipelines and co-usability performances of modules, Journal of Information Technologies 17 (2024) 95–107. URL: https://dergipark.org.tr/ en/pub/gazibtd/issue/84331/1399077. doi:10.17671/gazibtd.1399077

    2024 doi

  23. [31]

    Kasi, Yet another algorithm for pitch tracking (yaapt), 2002

    K. Kasi, Yet another algorithm for pitch tracking (yaapt), 2002

    2002

  24. [32]

    Eyben, M

    F. Eyben, M. Wöllmer, B. Schuller, Opensmile: the munich versatile and fast open-source audio feature extractor, in: Proceedings of the 18th ACM international conference on Multimedia, 2010, pp. 1459–1462

    2010

  25. [33]

    Sparck Jones, A statistical interpretation of term specificity and its application in retrieval, Journal of documentation 28 (1972) 11–21

    K. Sparck Jones, A statistical interpretation of term specificity and its application in retrieval, Journal of documentation 28 (1972) 11–21

    1972

  26. [34]

    J. Zhao, K. Liu, L. Xu, Sentiment analysis: Mining opinions, sentiments, and emotions, 2016

    2016

  27. [35]

    D. M. Argaw, S. Yoon, F. C. Heilbron, H. Deilamsalehy, T. Bui, Z. Wang, F. Dernoncourt, J. S. Chung, Scaling up video summarization pretraining with large language models, 2024. URL: https://arxiv.org/abs/2404.03398. arXiv:2404.03398

    2024 arXiv

  28. [36]

    Narasimhan, A

    M. Narasimhan, A. Nagrani, C. Sun, M. Rubinstein, T. Darrell, A. Rohrbach, C. Schmid, Tl; dw? summarizing instructional videos with task relevance and cross-modal saliency, in: European Conference on Computer Vision, Springer, 2022, pp. 540–557

    2022

Pith tools

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