Anchoring Trends: Mitigating Social Media Popularity Prediction Drift via Feature Clustering and Expansion

2025-07-26

AI Summary

Online video popularity prediction suffers from severe predictive drift: models exhibit sharp performance degradation on out-of-distribution (OOD) time periods due to rapidly evolving viral trends and user behavior. To address this, we propose a temporally robust multimodal popularity forecasting framework. Its core innovation is semantic anchor features—long-term stable semantic representations (e.g., content topics, audience attributes) generated by large language models (LLMs)—integrated with cross-temporal multimodal clustering to identify distribution-invariant patterns. We further jointly model these semantic anchors with statistical time-series features. Experiments on real-world social media data demonstrate substantial improvements in OOD prediction accuracy (+12.7% reduction in MAE) and model stability, effectively mitigating temporal distribution shift. Our approach establishes a transferable, generalizable paradigm for long-horizon popularity forecasting under dynamic environmental conditions.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Predicting online video popularity faces a critical challenge: prediction drift, where models trained on historical data rapidly degrade due to evolving viral trends and user behaviors. To address this temporal distribution shift, we propose an Anchored Multi-modal Clustering and Feature Generation (AMCFG) framework that discovers temporally-invariant patterns across data distributions. Our approach employs multi-modal clustering to reveal content structure, then leverages Large Language Models (LLMs) to generate semantic Anchor Features, such as audience demographics, content themes, and engagement patterns that transcend superficial trend variations. These semantic anchors, combined with cluster-derived statistical features, enable prediction based on stable principles rather than ephemeral signals. Experiments demonstrate that AMCFG significantly enhances both predictive accuracy and temporal robustness, achieving superior performance on out-of-distribution data and providing a viable solution for real-world video popularity prediction.

Problem

• Addressing prediction drift in social media popularity forecasting
• Identifying temporally-invariant patterns across evolving data distributions
• Enhancing accuracy and robustness in video popularity prediction

Innovation

• Multi-modal clustering for content structure
• LLM-generated semantic Anchor Features
• Cluster-derived statistical features for stability