AI RESEARCH
Predicting Post Virality with Temporal Cross-Attention over Trend Signals
arXiv CS.LG
•
ArXi:2605.02358v1 Announce Type: new Current models for predicting social media virality rely heavily on static textual and structural features, effectively ignoring the highly dynamic nature of trend signals. We study whether real-world attention signals can improve the prediction of social-media virality beyond what post text alone reveals. We