Predicting social media post popularity using multimodal deep learning: Insights from content features

Yi Wang and Aoyu Zeng

PLOS ONE, 2026, vol. 21, issue 8, 1-21

Abstract: This study predicts social media post popularity by examining the associations between multimodal, unstructured content features and observed engagement outcomes and by developing scalable, data-driven tools for social media content assessment. Grounded in Diffusion of Innovations, we conceptualize post popularity and the mechanisms of user engagement, extract visual and textual features from real-world social media data, validate their relationships with popularity, and build two deep learning models—one estimating numerical popularity and one predicting popularity levels. Results show that multimodal features, including visual and textual attributes, are significantly associated with popularity, and both models demonstrate useful predictive performance within the evaluated data setting. We extend the concept of post popularity by proposing a staged classification framework that groups posts into four levels based on aggregated engagement outcomes and demonstrate how multimodal feature analysis coupled with deep learning–based prediction offers new theoretical insight into content dissemination while yielding scalable tools for optimization. Practically, our approach enables marketers to assess popularity at both quantitative and categorical levels, supporting informed content planning, targeted promotion, and resource allocation, and helping brands move from intuition-driven to data-driven strategies.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0356612 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 56612&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0356612

DOI: 10.1371/journal.pone.0356612

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().