Pith. sign in

REVIEW 4 major objections 6 minor 59 references

Sentiment and Hashtag-aware Attentive Deep Neural Network for Multimodal Post Popularity Prediction

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A deep network that uses hashtags to guide attention over text and images, plus face-derived demographics and hashtag sentiment, predicts social media post popularity substantially better than seven existing methods on two real-world…

desk verdict Competent engineering with a genuinely new demographic feature, but the headline result is undermined by outcome-leaking features and no code/data. read the letter →

arxiv 2412.10737 v1 pith:XU2RBBD2 submitted 2024-12-14 cs.IR cs.SI

Shubhi Bansal , Mohit Kumar , Chandravardhan Singh Raghaw , Nagendra Kumar This is my paper · ORCID
classification cs.IRcs.SI
keywords multimodalpopularitypredictionhashtag-guidedattentionvisualdemographicshashtagsentimentsocialmediaanalysisdeepneuralnetworkfusionpost
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

Social media posts carry text, images, and hashtags, and predicting which ones will attract attention is hard because existing models mostly look at content alone. This paper proposes NARRATOR, a deep network that adds three underused signals: demographic traits (age, gender, race, emotion) read from faces in the image, sentiment expressed through hashtags, and an attention mechanism that lets hashtags steer which parts of the text and image matter most. The authors claim that NARRATOR beats seven state-of-the-art baselines on two real-world Flickr datasets, with the largest reported gains in mean squared error (1.196 versus 1.711 on TPIC, and 2.022 versus 4.211 on SMP). If that holds, it would show that hashtags are not just topical labels but contextual guides that connect visual and textual content to audience interests, and that face-derived demographics carry predictive signal beyond user metadata. The paper also reports ablation studies indicating each added component contributes to the gain.

What carries the argument

The load-bearing piece is the hashtag-guided attention mechanism: hashtags are embedded with BERT, then used to compute attention weights over the textual feature matrix (BERT embeddings passed through an LSTM) and the visual feature matrix (VGG19 regional features), so the attended content vector $\tilde{c}$ is the sum of hashtag-attended text and image representations. Around this sit two other novel inputs: a demographic feature vector $\{g,a,e,r\}$ (gender, age, emotion, race predicted by a face-analysis model) and a 10-dimensional sentiment vector built by concatenating CoreNLP sentiment scores for the caption and for the hashtags. All feature vectors are passed through separate CNN towers and fused in a deep feed-forward network that outputs the popularity score.

What would settle it

Retrain NARRATOR on the same TPIC and SMP splits with comment count and post duration removed from the social feature vector, and compare MSE against the reported values; if the model still keeps its margin over the next-best baselines, the gain is not mainly leakage. A second check is to train a variant using only comment count and duration plus the cheapest available features, and see how much of the reported improvement it recovers.

Watch

Extended reading notes

Core claim

The paper's central claim is that the popularity of a multimodal post can be predicted more accurately by explicitly modeling what hashtags do to the interpretation of text and images, and by adding two previously neglected modalities: visual demographics and hashtag sentiment. On the TPIC2017 and SMP datasets, NARRATOR reports MSE of 1.196 and 2.022 respectively, outperforming all seven compared methods; the next-best baselines are VSCNN with 1.711 on TPIC and TweetGage with 4.211 on SMP. Ablation experiments show that removing hashtag sentiment and demographics together raises MSE by about 14 percent on both datasets, and replacing hashtag-guided attention with self-, cross-, or no attention increases error, supporting the claim that the new components are responsible for the improvement.

Load-bearing premise

The evaluation assumes that all input features, including comment count and post duration, are known before the popularity label that the model is supposed to predict, so the reported gains may partly reflect these outcome-correlated variables rather than genuine forecasting skill.

Editorial extensions

If this is right

  • On the two Flickr-derived datasets, NARRATOR reports MSE reductions of roughly 30 to 78 percent over seven baselines, implying the added modalities carry signal not captured by content-only or metadata-only models.
  • Ablation studies attribute part of the gain to visual demographics and hashtag sentiment, and the attention study attributes part to hashtag guidance over self-, cross-, and parallel co-attention.
  • The learned attention weights could be inspected to show which image regions and caption words a hashtag highlights, making the predictions more interpretable for users and content creators.
  • The model is computationally heavier at training than most baselines, but its inference time is low enough that the authors describe it as suitable for near-real-time predictions.

Reading between the lines

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

  • A clean test would retrain NARRATOR with comment count and post duration held out, because those features are only observable after a post has been online and may inflate the reported forecasting accuracy.
  • The hashtag-guided attention idea could transfer to other multimodal tasks where tags or labels provide context, such as content recommendation or content moderation, though that would need separate validation.
  • Face-derived demographics carry privacy and bias risks; if this approach moves toward deployment, consent, fairness, and generalization across platforms would need explicit examination.
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 / 6 minor

Summary. The paper proposes NARRATOR, a multimodal deep neural network for predicting the popularity of social media posts, formulated as a regression to log-normalized view counts. The model combines BERT/LSTM text features, VGG19 visual features, DeepFace-derived demographic attributes, topic and graph-structure hashtag embeddings, CoreNLP sentiment features for captions and hashtags, user/post metadata, and a novel hashtag-guided attention mechanism that reweights text and image features using hashtag context. These representations are passed through per-modality CNNs, concatenated with the attention output, and fed to a 12-layer feedforward network. The empirical claim is that NARRATOR outperforms seven state-of-the-art baselines by a large margin on the TPIC and SMP datasets (Table 1). The paper also reports feature ablations, attention-mechanism comparisons, correlation analyses, feature ranking, and computational cost.

Significance. If the empirical claim were sound, the work would be a useful contribution to multimodal popularity prediction by demonstrating that hashtag-guided attention, visual demographics, and hashtag sentiment improve forecasting accuracy, with practical implications for recommendation, advertising, and trend analysis. The paper is honest about computational costs and limitations, and it provides ablations that isolate the proposed components. However, the evaluation is compromised by the inclusion of post-hoc outcome variables (comment count and post duration) as input features, by single-run metrics without variance estimates, and by comparisons on filtered subsets whose relationship to the baselines' original evaluation data is unspecified. As presented, the reported margins over baselines cannot be distinguished from outcome leakage, so the central claim is unsupported. The proposed mechanism is interesting, but its benefit is not demonstrated by the current experiments.

major comments (4)
  1. [§4.1.6, §4.3 (Eq. 14), Table 1] Comment Count (Post Metadata item e) and Post Duration (Time item d) are included in the 85-dimensional social feature vector f^s_i, which is PCA-reduced to 6 dimensions and fed through the CNN stream into the fused representation M_i and the final predictor. The label is the log-normalized view count, and the paper frames the task as forecasting popularity 'early on' (Section 1); comment count is a contemporaneous engagement outcome and post duration is the length of the observation window over which views accumulate, so neither is available at the time the prediction is supposed to be made. The large margins over the best baselines in Table 1 (TPIC MSE 1.196 vs. 1.711; SMP MSE 2.022 vs. 4.211) are therefore consistent with outcome leakage rather than predictive skill. No ablation without these two features, no temporal split, and no error bars are provided, so the central claim of the paper is unsupported as presented.
  2. [§5.1.1 (Dataset Preprocessing) and §5.2.7] The evaluation uses filtered subsets (21,000 SMP and 11,000 TPIC posts) restricted to posts that contain hashtags, titles, and faces. The paper does not state whether the baselines in Table 1 were re-run on the same filtered subsets with the same train/validation/test split; if their numbers are taken from the original papers, the comparison is against different data and the reported margins (e.g., 30-70% relative improvements) may reflect the changed evaluation set rather than model quality. The test sets are also small (about 2,100 and 1,100 posts, respectively) for a model with 34.9M parameters (Section 5.2.8), which makes the absence of variance estimates particularly concerning.
  3. [§5.2.1 (Table 1) and §5.2.4] All effectiveness results are single-run point estimates without standard deviations, confidence intervals, or significance tests. The statistical analysis reports SRCC values of 0.885 and 0.8773 and PCC values of 0.8355 and 0.8409 with p ≤ 0.05, but gives no test statistic, null hypothesis, sample size, or multiple-comparison correction. As a result, the claimed significant margin over baselines cannot be evaluated for statistical reliability, which is load-bearing for the empirical claim.
  4. [§4.2 (Algorithm 1)] Algorithm 1 has inconsistent notation: line 2 uses V[t] as a learnable parameter while lines 4-5 use V[i] as the image feature matrix from Eq. (6); the text defines U^t, V^t, U^i, V^i as learnable parameters. This makes the exact computation of the hashtag-guided attention mechanism ambiguous and impedes reproduction of the central methodological contribution. Additionally, Section 4.1.3 states the demographic feature vector has dimension 116, while Section 4.3 says it has dimension four, so the stated merged vector size of 27104 in Section 4.3 cannot be verified.
minor comments (6)
  1. [§4.4] The text says the output layer 'directly predict[s] the continuous star count'; this should be the popularity score or view count. Equation (20) also says the objective is 'maximizing the Mean Squared Error (MSE) cost function' after defining MSE as the loss to be minimized.
  2. [§5.2.6 (Table 5)] The entries in Table 5 are garbled (e.g., '1 .362(5)/0.867(6) 5.5'), making the feature ranking difficult to read; please reformat the table.
  3. [§6.2] Parallel co-attention is cited as [25], but [25] (Zhang et al., hashtag recommendation) is not the source of parallel co-attention used earlier in the paper; the relevant citation appears to be [43] (Bansal et al.).
  4. [§5.2.7] It is stated that the embedding dimension D=768 is set uniformly for all comparative methods, but it is not explained how the baselines (e.g., TweetGage or HashPop) are adapted to this setting; please clarify whether all baselines were re-implemented and re-run under identical conditions.
  5. [§4.1.2] Equation (7) reuses the symbol v^k_i from Eq. (6) but now denotes the projected feature vector in R^D; using a different symbol would avoid confusion.
  6. [§6.4.2] The practical implication 'Combating Misinformation' claims that predicting popularity early can help identify harmful content, but popularity is not the same as harmfulness; this implication needs qualification.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the claimed gains are benchmarked against external datasets and external baselines, and no prediction is defined by or fitted to its own target.

full rationale

The paper's central claim, that NARRATOR outperforms existing methods on TPIC and SMP, is established by comparing an independently trained model against external baselines on held-out test partitions of two public benchmark datasets. The target (log-normalized view count) is external to the model and is never used to define any input feature. The proposed components (visual demographics, hashtag sentiment, hashtag-guided attention) are model inputs or architectural mechanisms, not functions of the ground-truth popularity labels, so the derivation does not reduce to its own inputs. The hashtag-guided attention mechanism is validated by ablation against no-attention, self-attention, cross-attention, and parallel co-attention variants; this is an empirical comparison, not a tautology. The paper contains minor self-citations ([42], [43]) in related work, but these are background references and are not load-bearing for the claimed result. The inclusion of post-hoc features such as Comment Count and Post Duration is a potential evaluation-leakage concern (a correctness risk), not a circularity of the kind where a prediction is equivalent to a fitted input by construction. No uniqueness theorem is invoked, no fitted quantity is renamed as a prediction, and no known result is repackaged as new under a different name without independent empirical support. Therefore, no specific circular step can be quoted and exhibited, and the appropriate finding is no significant circularity.

Assumptions & free parameters 11 free parameters · 5 assumptions · 0 invented entities

The model's predictive claim rests on the validity of several pretrained components and on an unjustified assumption that post-hoc social features are available at prediction time. The latter is the dominant risk to the reported results.

free parameters (11)
  • Embedding dimension D = 768
    Used for text, image, and attention projections; chosen by the authors and forced to be equal for all baselines.
  • Maximum text length M = 15
    Truncates and pads captions; chosen without reported sensitivity analysis.
  • Maximum hashtag count L = 60
    Pads hashtag sequences to length 60; chosen without reported sensitivity analysis.
  • Attention units A = 768
    Number of attention units in hashtag-guided attention.
  • PCA components for social features = 6
    Reduces the social feature vector from 85 to 6 dimensions; retained variance is not reported.
  • Learning rate = 0.0001
    Hyperparameter from tuning; no search range reported.
  • Batch size = 20
    Selected for training stability.
  • Training epochs and early stopping = 30 max, patience 5
    Early stopping rule chosen by hand.
  • Dropout rate = 0.2
    Applied after each layer.
  • DNN architecture = 12 FC layers, widths 13552, 6776, ..., 1
    Final predictor architecture chosen by hand without ablation.
  • CNN layer configuration for modality branches = 3 Conv1D layers per branch, filter sizes and kernels not reported
    Conv1D layers are used for social, demographic, hashtag, and sentiment features; the paper does not state filter counts or kernel sizes.
assumptions (5)
  • domain assumption Off-the-shelf feature extractors, BERT, VGG19, DeepFace, BERTopic, GraphSAGE, CoreNLP, produce valid representations for the target social media posts.
    Section 4.1 uses these tools without validation on the target datasets.
  • domain assumption DeepFace demographic outputs, age, gender, race, emotion, are accurate enough to add predictive signal.
    Section 4.1.3; no quality control for face detection errors is reported, and the authors note lighting and occlusion limits in Section 6.3.
  • domain assumption Stanford CoreNLP sentence-level sentiment applies meaningfully to hashtag strings.
    Section 4.1.5 treats hashtags as sentences without evaluation.
  • ad hoc to paper Comment count and post duration are legitimate input features for early popularity prediction.
    Section 4.1.6 lists them as features, but they become known only after the post has accumulated engagement, so they can leak the label.
  • domain assumption Filtering to posts with faces, hashtags, and titles preserves a representative evaluation set.
    Section 5.1.1 reduces SMP to 21,000 and TPIC to 11,000 samples; the selected subset may not represent general posts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sentiment and Hashtag-aware Attentive Deep Neural Network for Multimodal Post Popularity Prediction." pith.science (2026). https://pith.science/paper/XU2RBBD2

@misc{pith2026241210737,
  author       = {Pith},
  title        = {Pith review of: Sentiment and Hashtag-aware Attentive Deep Neural Network for Multimodal Post Popularity Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XU2RBBD2}},
  note         = {Machine review of arXiv:2412.10737}
}
read the original abstract

Social media users articulate their opinions on a broad spectrum of subjects and share their experiences through posts comprising multiple modes of expression, leading to a notable surge in such multimodal content on social media platforms. Nonetheless, accurately forecasting the popularity of these posts presents a considerable challenge. Prevailing methodologies primarily center on the content itself, thereby overlooking the wealth of information encapsulated within alternative modalities such as visual demographics, sentiments conveyed through hashtags and adequately modeling the intricate relationships among hashtags, texts, and accompanying images. This oversight limits the ability to capture emotional connection and audience relevance, significantly influencing post popularity. To address these limitations, we propose a seNtiment and hAshtag-aware attentive deep neuRal netwoRk for multimodAl posT pOpularity pRediction, herein referred to as NARRATOR that extracts visual demographics from faces appearing in images and discerns sentiment from hashtag usage, providing a more comprehensive understanding of the factors influencing post popularity Moreover, we introduce a hashtag-guided attention mechanism that leverages hashtags as navigational cues, guiding the models focus toward the most pertinent features of textual and visual modalities, thus aligning with target audience interests and broader social media context. Experimental results demonstrate that NARRATOR outperforms existing methods by a significant margin on two real-world datasets. Furthermore, ablation studies underscore the efficacy of integrating visual demographics, sentiment analysis of hashtags, and hashtag-guided attention mechanisms in enhancing the performance of post popularity prediction, thereby facilitating increased audience relevance, emotional engagement, and aesthetic appeal.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 25 canonical work pages

  1. [1]

    Telematics and Informatics 47, 101303 (2020) https://doi.org/10.1016/j

    Shen, F., Xia, C., Skoric, M.: Examining the roles of social media and alternative media in social movement participation: A study of hong kong’s umbrella move- ment. Telematics and Informatics 47, 101303 (2020) https://doi.org/10.1016/j. tele.2019.101303

    arXiv 2020

  2. [2]

    Anderson, M., Brook, A.: Social Media Use in 2021 — Pew Research Center (2021)

    work page 2021

  3. [4]

    In: 38 Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Wu, B., Mei, T., Cheng, W.-H., Zhang, Y.: Unfolding temporal dynamics: Pre- dicting social media popularity using multi-scale temporal decomposition. In: 38 Proceedings of the AAAI Conference on Artificial Intelligence, vol. 30 (2016). https://doi.org/10.1609/aaai.v30i1.9970

    work page doi:10.1609/aaai.v30i1.9970 2016

  4. [5]

    IEEE Internet Computing 14(3), 42–49 (2010) https://doi.org/10.1109/MIC.2010.73

    Gon¸ calves, M.A., Almeida, J.M., Santos, L.G., Laender, A.H., Almeida, V.: On popularity in the blogosphere. IEEE Internet Computing 14(3), 42–49 (2010) https://doi.org/10.1109/MIC.2010.73

    work page doi:10.1109/mic.2010.73 2010

  5. [6]

    International Journal of Geographical Information Science 27(4), 662–684 (2013) https://doi.org/10.1080/13658816.2012.696649

    Majid, A., Chen, L., Chen, G., Mirza, H.T., Hussain, I., Woodward, J.: A context- aware personalized travel recommendation system based on geotagged social media data mining. International Journal of Geographical Information Science 27(4), 662–684 (2013) https://doi.org/10.1080/13658816.2012.696649

    arXiv 2013

  6. [7]

    In: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp

    Li, C., Lu, Y., Mei, Q., Wang, D., Pandey, S.: Click-through prediction for adver- tising in twitter timeline. In: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1959–1968 (2015). https://doi.org/10.1145/2783258.2788582

    arXiv 2015

  7. [8]

    Google Patents

    Aven, B.L., Burgess, D.A., Haynes, J.F., Merino, J.R., Moore, P.C.: Using product and social network data to improve online advertising. Google Patents. US Patent 8,843,406 (2014)

    work page 2014

  8. [9]

    IEEE Transactions on multimedia 15(6), 1255–1267 (2013) https://doi.org/10.1109/TMM.2013.2265079

    Roy, S.D., Mei, T., Zeng, W., Li, S.: Towards cross-domain learning for social video popularity prediction. IEEE Transactions on multimedia 15(6), 1255–1267 (2013) https://doi.org/10.1109/TMM.2013.2265079

    work page arXiv 2013

Show all 59 references
  1. [10]

    In: Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III 14, pp

    Gan, C., Sun, C., Duan, L., Gong, B.: Webly-supervised video recognition by mutually voting for relevant web images and web video frames. In: Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III 14, pp. 849...

    2016 doi

  2. [11]

    International Journal of Distributed Sensor Networks 11(9), 907169 (2015) https://doi.org/10

    Kim, W., Won, J.H., Park, S., Kang, J.: Demand forecasting models for medicines through wireless sensor networks data and topic trend analysis. International Journal of Distributed Sensor Networks 11(9), 907169 (2015) https://doi.org/10. 1155/2015/907169

    2015

  3. [12]

    In: 2021 International Conference on Machine Learning and Intelligent Systems En- gineering (MLISE), pp

    Wang, J., Xu, B., Zu, Y.: Deep learning for aspect-based sentiment analysis. In: 2021 International Conference on Machine Learning and Intelligent Systems En- gineering (MLISE), pp. 267–271 (2021). https://doi.org/10.1109/MLISE54096. 2021.00056 . IEEE

    2021

  4. [13]

    Journal of Innovation & Knowledge 6(2), 92–102 (2021) https://doi.org/10.1016/j.jik.2020.08.001

    Saura, J.R.: Using data sciences in digital marketing: Framework, methods, and performance metrics. Journal of Innovation & Knowledge 6(2), 92–102 (2021) https://doi.org/10.1016/j.jik.2020.08.001

    2021 doi

  5. [14]

    International Journal of Information Management 60, 102331 (2021) https://doi.org/10.1016/j.ijinfomgt.2021.102331

    Saura, J.R., Ribeiro-Soriano, D., Palacios-Marqu´ es, D.: From user-generated data to data-driven innovation: A research agenda to understand user privacy in digital 39 markets. International Journal of Information Management 60, 102331 (2021) https://doi.org/10.1016/j.ijinfom...

    2021

  6. [15]

    Technological Forecasting and Social Change 167, 120681 (2021) https://doi.org/10.1016/j.techfore.2021.120681

    Ribeiro-Navarrete, S., Saura, J.R., Palacios-Marqu´ es, D.: Towards a new era of mass data collection: Assessing pandemic surveillance technologies to preserve user privacy. Technological Forecasting and Social Change 167, 120681 (2021) https://doi.org/10.1016/j.techfore.2021.120681

    2021

  7. [16]

    In: MM 2020 - Proceedings of the 28th ACM International Conference on Multimedia (2020)

    Xu, K., Lin, Z., Zhao, J., Shi, P., Deng, W., Wang, H.: Multimodal deep learning for social media popularity prediction with attention mechanism. In: MM 2020 - Proceedings of the 28th ACM International Conference on Multimedia (2020). https://doi.org/10.1145/3394171.3416274

    2020

  8. [17]

    IEEE Access 10 (2022) https://doi.org/10.1109/ACCESS.2021.3136552

    Lin, H.H., Lin, J.D., Ople, J.J.M., Chen, J.C., Hua, K.L.: Social media popular- ity prediction based on multi-modal self-attention mechanisms. IEEE Access 10 (2022) https://doi.org/10.1109/ACCESS.2021.3136552

    2022

  9. [18]

    Nguyen, M.-T., Le, D.H., Nakajima, T., Yoshimi, M., Thoai, N.: Attention-based neural network: A novel approach for predicting the popularity of online content. In: 2019 IEEE 21st International Conference on High Performance Computing and Communications; IEEE 17th Internationa...

    2019

  10. [19]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Liao, D., Xu, J., Li, G., Huang, W., Liu, W., Li, J.: Popularity prediction on online articles with deep fusion of temporal process and content features. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 33, pp. 200–207 (2019). https://doi.org/10.1609/aaa...

    2019 doi

  11. [21]

    In: 2018 IEEE International Con- ference on Big Data (big Data), pp

    Zhang, Z., Chen, T., Zhou, Z., Li, J., Luo, J.: How to become instagram famous: Post popularity prediction with dual-attention. In: 2018 IEEE International Con- ference on Big Data (big Data), pp. 2383–2392 (2018). https://doi.org/10.1109/ BigData.2018.8622461 . IEEE

    2018

  12. [22]

    In: The Web Conference 2018 - Proceedings of the World Wide Web Conference, WWW 2018 (2018)

    Zhang, W., Wang, W., Wang, J., Zha, H.: User-guided hierarchical attention net- work for multi-modal social image popularity prediction. In: The Web Conference 2018 - Proceedings of the World Wide Web Conference, WWW 2018 (2018). https://doi.org/10.1145/3178876.3186026

    2018

  13. [24]

    Caleffi, P.-M.: The ‘hashtag’: A new word or a new rule? SKASE Journal of Theoretical Linguistics 12 (2015) https://doi.org/10.24093/awej/call6.6

    2015 doi

  14. [25]

    In: Proceedings of the AAAI Conference on Artificial Intelligence (2019)

    Zhang, S., Yao, Y., Xu, F., Tong, H., Yan, X., Lu, J.: Hashtag recommendation for photo sharing services. In: Proceedings of the AAAI Conference on Artificial Intelligence (2019). https://doi.org/10.1609/aaai.v33i01.33015805

    2019 doi

  15. [26]

    IEEE Transactions on Cognitive and Developmental Systems 13 (2021) https://doi.org/10.1109/TCDS

    Abousaleh, F.S., Cheng, W.H., Yu, N.H., Tsao, Y.: Multimodal deep learning framework for image popularity prediction on social media. IEEE Transactions on Cognitive and Developmental Systems 13 (2021) https://doi.org/10.1109/TCDS. 2020.3036690

    2021

  16. [27]

    In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (2014)

    Bakhshi, S., Shamma, D.A., Gilbert, E.: Faces engage us: Photos with faces at- tract more likes and comments on instagram. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (2014). https://doi.org/10. 1145/2556288.2557403

    2014

  17. [28]

    In: Proceedings of the 23rd ACM International Conference on Multimedia (2015)

    Gelli, F., Uricchio, T., Bertini, M., Bimbo, A.D., Chang, S.F.: Image popularity prediction in social media using sentiment and context features. In: Proceedings of the 23rd ACM International Conference on Multimedia (2015). https://doi. org/10.1145/2733373.2806361

    2015

  18. [29]

    In: Proceedings - IEEE International Con- ference on Data Mining, ICDM, vol

    Li, J., Gao, Y., Gao, X., Shi, Y., Chen, G.: Senti2pop: Sentiment-aware topic popularity prediction on social media. In: Proceedings - IEEE International Con- ference on Data Mining, ICDM, vol. 2019-November (2019). https://doi.org/10. 1109/ICDM.2019.00143

    2019

  19. [31]

    IEEE Access 8, 78252–78264 (2020) https://doi.org/ 10.1109/ACCESS.2020.2989473

    Yang, C., Wang, X., Jiang, B.: Sentiment enhanced multi-modal hashtag recom- mendation for micro-videos. IEEE Access 8, 78252–78264 (2020) https://doi.org/ 10.1109/ACCESS.2020.2989473

    2020

  20. [32]

    In: Calzolari, N., B´ echet, F., Blache, P., Choukri, K., Cieri, C., Declerck, T., Goggi, S., Isahara, H., Maegaard, B., Mariani, J., Mazo, H., Odijk, J., Piperidis, S

    Liao, Y.Y.: Leveraging hashtag networks for multimodal popularity prediction of Instagram posts. In: Calzolari, N., B´ echet, F., Blache, P., Choukri, K., Cieri, C., Declerck, T., Goggi, S., Isahara, H., Maegaard, B., Mariani, J., Mazo, H., Odijk, J., Piperidis, S. (eds.) Proc...

    2022

  21. [33]

    In: Proceedings of the 2023 ACM International Conference on Multimedia Retrieval, pp

    Arazzi, M., Cotogni, M., Nocera, A., Virgili, L.: Predicting tweet engagement with 41 graph neural networks. In: Proceedings of the 2023 ACM International Conference on Multimedia Retrieval, pp. 172–180 (2023). https://doi.org/10.1145/3591106. 3592294

    2023 doi

  22. [34]

    International Arab Journal of Information Technology 18 (2021) https: //doi.org/10.34028/iajit/18/1/10

    Purba, K.R., Asirvatham, D., Murugesan, R.K.: Instagram post popularity trend analysis and prediction using hashtag, image assessment, and user history fea- tures. International Arab Journal of Information Technology 18 (2021) https: //doi.org/10.34028/iajit/18/1/10

    2021 doi

  23. [35]

    10682 LNAI (2017)

    Kumar, N., Yadandla, A., Suryamukhi, K., Ranabothu, N., Boya, S., Singh, M.: Arousal prediction of news articles in social media, vol. 10682 LNAI (2017). https: //doi.org/10.1007/978-3-319-71928-3 30

    2017 doi

  24. [36]

    World Wide Web 22 (2019) https://doi

    Lin, Z., Huang, F., Li, Y., Yang, Z., Liu, W.: A layer-wise deep stacking model for social image popularity prediction. World Wide Web 22 (2019) https://doi. org/10.1007/s11280-018-0590-1

    2019 doi

  25. [37]

    Cao, Q., Shen, H., Gao, J., Wei, B., Cheng, X.: Popularity prediction on social platforms with coupled graph neural networks. (2020). https://doi.org/10.1145/ 3336191.3371834

    2020

  26. [38]

    International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems31 (2023) https: //doi.org/10.1142/S0218488523500150

    Mannepalli, K., Singh, S.P., Kolli, C.S., Raj, S., Bojja, G.R., Rajakumar, B.R., Binu, D.: Popularity prediction model with context, time and user sentiment information: An optimization assisted deep learning technique. International Journal of Uncertainty, Fuzziness and Knowl...

    2023 doi

  27. [39]

    In: MM 2022 - Proceedings of the 30th ACM International Conference on Multimedia, pp

    Tan, Y., Liu, F., Li, B., Zhang, Z., Zhang, B.: An efficient multi-view multimodal data processing framework for social media popularity prediction. In: MM 2022 - Proceedings of the 30th ACM International Conference on Multimedia, pp. 7200–7204 (2022). https://doi.org/10.1145/...

    2022

  28. [40]

    Social Semiotics 25(3), 274–291 (2015) https://doi.org/10.1080/10350330.2014.996948

    Zappavigna, M.: Searchable talk: The linguistic functions of hashtags. Social Semiotics 25(3), 274–291 (2015) https://doi.org/10.1080/10350330.2014.996948

    2015

  29. [41]

    In: IJCAI, vol

    Liu, J., He, Z., Huang, Y.: Hashtag2vec: Learning hashtag representation with relational hierarchical embedding model. In: IJCAI, vol. 2018-July (2018). https: //doi.org/10.24963/ijcai.2018/480

    2018 doi

  30. [42]

    Social Network Analysis and Mining 13 (2023) https://doi.org/10.1007/s13278-023-01024-9

    Chakrabarti, P., Malvi, E., Bansal, S., Kumar, N.: Hashtag recommendation for enhancing the popularity of social media posts. Social Network Analysis and Mining 13 (2023) https://doi.org/10.1007/s13278-023-01024-9

    2023 doi

  31. [43]

    IEEE Transactions on Computational Social Systems 10 (2023) https://doi.org/10.1109/TCSS.2022.3184307 42

    Bansal, S., Gowda, K., Kumar, N.: A hybrid deep neural network for multimodal personalized hashtag recommendation. IEEE Transactions on Computational Social Systems 10 (2023) https://doi.org/10.1109/TCSS.2022.3184307 42

    2023

  32. [44]

    Computer Speech and Language 80 (2023) https://doi.org/10.1016/j.csl.2023.101490

    Wang, J., Yang, S., Zhao, H., Yang, Y.: Social media popularity prediction with multimodal hierarchical fusion model. Computer Speech and Language 80 (2023) https://doi.org/10.1016/j.csl.2023.101490

    2023

  33. [45]

    In: Burstein, J., Doran, C., Solorio, T

    Devlin, J., Chang, M.-W., Lee, K., Toutanova, K.: BERT: Pre-training of deep bidirectional transformers for language understanding. In: Burstein, J., Doran, C., Solorio, T. (eds.) Proceedings of the 2019 Conference of the North Ameri- can Chapter of the Association for Computa...

    2019

  34. [46]

    Advances in Neural Information Processing Systems (2013)

    Mikolov, T., Sutskever, I., Chen, K., Corrado, G., Dean, J.: Distributed repre- sentations of words and phrases and their compositionality. Advances in Neural Information Processing Systems (2013)

    2013

  35. [47]

    In: 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings (2015)

    Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. In: 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings (2015)

    2015

  36. [48]

    In: 2009 IEEE Conference on Computer Vision and Pattern Recognition (2010)

    Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., Fei-Fei, L.: Imagenet: A large- scale hierarchical image database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition (2010). https://doi.org/10.1109/cvpr.2009.5206848

    2010

  37. [49]

    In: 2021 International Conference on Engineering and Emerging Tech- nologies (ICEET) (2021)

    Serengil, S.I., Ozpinar, A.: Hyperextended lightface: A facial attribute analysis framework. In: 2021 International Conference on Engineering and Emerging Tech- nologies (ICEET) (2021). https://doi.org/10.1109/ICEET53442.2021.9659697

    2021

  38. [50]

    https://arxiv.org/abs/2203.05794 (2022)

    Grootendorst, M.: Bertopic: Neural topic modeling with a class-based tf-idf procedure. https://arxiv.org/abs/2203.05794 (2022)

    2022 arXiv

  39. [51]

    The Journal of Open Source Software 2 (2017) https://doi.org/10.21105/joss

    McInnes, L., Healy, J., Astels, S.: hdbscan: Hierarchical density based clustering. The Journal of Open Source Software 2 (2017) https://doi.org/10.21105/joss. 00205

    2017 doi

  40. [52]

    Journal of Open Source Software 3 (2018) https: //doi.org/10.21105/joss.00861

    McInnes, L., Healy, J., Saul, N., Großberger, L.: Umap: Uniform manifold ap- proximation and projection. Journal of Open Source Software 3 (2018) https: //doi.org/10.21105/joss.00861

    2018 doi

  41. [53]

    In: Neural Information Processing Systems, vol

    Hamilton, W.L., Ying, R., Leskovec, J.: Inductive representation learning on large graphs. In: Neural Information Processing Systems, vol. 2017-December (2017)

    2017

  42. [54]

    In: Proceedings of 52nd Annual Meeting of the Association for Computational Linguistics: System Demonstrations, vol

    Manning, C.D., Surdeanu, M., Bauer, J., Finkel, J., Bethard, S.J., McClosky, D.: The stanford corenlp natural language processing toolkit. In: Proceedings of 52nd Annual Meeting of the Association for Computational Linguistics: System Demonstrations, vol. 2014-June (2014). htt...

    2014 doi

  43. [55]

    In: Proceedings of the International AAAI Con- ference on Web and Social Media (2014)

    Hutto, C.J., Gilbert, E.: Vader: A parsimonious rule-based model for sentiment analysis of social media text. In: Proceedings of the International AAAI Con- ference on Web and Social Media (2014). https://doi.org/10.1609/icwsm.v8i1. 14550

    2014 doi

  44. [56]

    Sensors (Switzerland) 17 (2017) https://doi.org/10.3390/s17030631

    Aloufi, S., Zhu, S., Saddik, A.E.: On the prediction of flickr image popularity by analyzing heterogeneous social sensory data. Sensors (Switzerland) 17 (2017) https://doi.org/10.3390/s17030631

    2017 doi

  45. [57]

    https://doi.org/10.1098/rsta.2015.0202

    Jollife, I.T., Cadima, J.: Principal component analysis: A review and recent developments (2016). https://doi.org/10.1098/rsta.2015.0202

    2016

  46. [58]

    journal of machine learning research

    Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R.: Dropout: a simple way to prevent neural networks from overfitting. journal of machine learning research. Journal of Machine Learning Research 15 (2014)

    2014

  47. [60]

    In: IJCAI International Joint Conference on Artificial Intelligence, vol

    Wu, B., Cheng, W.H., Zhang, Y., Huang, Q., Li, J., Mei, T.: Sequential pre- diction of social media popularity with deep temporal context networks. In: IJCAI International Joint Conference on Artificial Intelligence, vol. 0 (2017). https://doi.org/10.24963/ijcai.2017/427

    2017 doi

  48. [61]

    In: Proceedings of the 27th ACM International Conference on Multimedia (2019)

    Ding, K., Wang, R., Wang, S.: Social media popularity prediction: A multiple feature fusion approach with deep neural networks. In: Proceedings of the 27th ACM International Conference on Multimedia (2019). https://doi.org/10.1145/ 3343031.3356062

    2019

  49. [62]

    In: Proceedings of the 31st ACM International Con- ference on Multimedia, pp

    Gradient boost tree network based on extensive feature analysis for popularity prediction of social posts. In: Proceedings of the 31st ACM International Con- ference on Multimedia, pp. 9451–9455 (2023). https://doi.org/10.1145/3581783. 3612843

    2023 doi

  50. [63]

    In: Proceedings of the 31st ACM International Conference on Multimedia, pp

    Mao, S., Xi, W., Yu, L., L¨ u, G., Xing, X., Zhou, X., Wan, W.: Enhanced catboost with stacking features for social media prediction. In: Proceedings of the 31st ACM International Conference on Multimedia, pp. 9430–9435 (2023). https:// doi.org/10.1145/3581783.3612839

    2023

  51. [64]

    The American Journal of Psychology 15 (1904) https://doi.org/10.2307/1412159 44

    Spearman, C.: The proof and measurement of association between two things. The American Journal of Psychology 15 (1904) https://doi.org/10.2307/1412159 44

    1904 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.