@inproceedings{wang-etal-2026-multimodal,
title = "A {M}ulti{M}odal Autoregressive Transformer Framework for Social Popularity Prediction",
author = "Wang, Mingzhu and
Zhou, Dong and
Zhou, Zeliang",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1094/",
doi = "10.18653/2026.findings-emnlp.1094",
pages = "23162--23175",
ISBN = "979-8-89176-462-0",
abstract = "Understanding and predicting the temporal popularity of social media content is critical for digital marketing and recommendation systems. Existing methods often rely on static feature concatenation, struggling to capture complex multimodal interactions and strict sequential dependencies in long-term popularity evolution, frequently underestimating viral content. To address these limitations, we propose a MultiModal Autoregressive Transformer Framework (MMATF). Our framework extracts and refines key representations from video covers, texts, category tags, and numerical statistics. A dynamic modality interaction module adaptively fuses these inputs to capture evolving content dynamics, establishing a solid foundation for temporal relevance. Furthermore, an autoregressive temporal encoding module strictly prevents future information leakage. Extensive experiments on SMTPD and ICIP datasets demonstrate MMATF significantly outperforms advanced baselines across 7-, 14-, and 30-day horizons. Specifically, compared to the strongest baseline, our model reduces Mean Absolute Error by 3.8{\%} and 4.3{\%}, and improves Spearman{'}s Rank Correlation by 0.3{\%} and 0.4{\%} for 30-day predictions on SMTPD and ICIP respectively, proving consistent long-term forecasting accuracy."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-multimodal">
<titleInfo>
<title>A MultiModal Autoregressive Transformer Framework for Social Popularity Prediction</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mingzhu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dong</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zeliang</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Understanding and predicting the temporal popularity of social media content is critical for digital marketing and recommendation systems. Existing methods often rely on static feature concatenation, struggling to capture complex multimodal interactions and strict sequential dependencies in long-term popularity evolution, frequently underestimating viral content. To address these limitations, we propose a MultiModal Autoregressive Transformer Framework (MMATF). Our framework extracts and refines key representations from video covers, texts, category tags, and numerical statistics. A dynamic modality interaction module adaptively fuses these inputs to capture evolving content dynamics, establishing a solid foundation for temporal relevance. Furthermore, an autoregressive temporal encoding module strictly prevents future information leakage. Extensive experiments on SMTPD and ICIP datasets demonstrate MMATF significantly outperforms advanced baselines across 7-, 14-, and 30-day horizons. Specifically, compared to the strongest baseline, our model reduces Mean Absolute Error by 3.8% and 4.3%, and improves Spearman’s Rank Correlation by 0.3% and 0.4% for 30-day predictions on SMTPD and ICIP respectively, proving consistent long-term forecasting accuracy.</abstract>
<identifier type="citekey">wang-etal-2026-multimodal</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1094</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1094/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>23162</start>
<end>23175</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A MultiModal Autoregressive Transformer Framework for Social Popularity Prediction
%A Wang, Mingzhu
%A Zhou, Dong
%A Zhou, Zeliang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F wang-etal-2026-multimodal
%X Understanding and predicting the temporal popularity of social media content is critical for digital marketing and recommendation systems. Existing methods often rely on static feature concatenation, struggling to capture complex multimodal interactions and strict sequential dependencies in long-term popularity evolution, frequently underestimating viral content. To address these limitations, we propose a MultiModal Autoregressive Transformer Framework (MMATF). Our framework extracts and refines key representations from video covers, texts, category tags, and numerical statistics. A dynamic modality interaction module adaptively fuses these inputs to capture evolving content dynamics, establishing a solid foundation for temporal relevance. Furthermore, an autoregressive temporal encoding module strictly prevents future information leakage. Extensive experiments on SMTPD and ICIP datasets demonstrate MMATF significantly outperforms advanced baselines across 7-, 14-, and 30-day horizons. Specifically, compared to the strongest baseline, our model reduces Mean Absolute Error by 3.8% and 4.3%, and improves Spearman’s Rank Correlation by 0.3% and 0.4% for 30-day predictions on SMTPD and ICIP respectively, proving consistent long-term forecasting accuracy.
%R 10.18653/2026.findings-emnlp.1094
%U https://aclanthology.org/2026.findings-emnlp.1094/
%U https://doi.org/10.18653/2026.findings-emnlp.1094
%P 23162-23175
Markdown (Informal)
[A MultiModal Autoregressive Transformer Framework for Social Popularity Prediction](https://aclanthology.org/2026.findings-emnlp.1094/) (Wang et al., Findings 2026)
ACL