Research map / Large language models / Retrieval and knowledge
Learning to Curate Context: Jointly Optimizing Retrieval and Prediction for Multimodal Social Media Popularity
Xovee Xu, Shuojun Lin, Fan Zhou, Jingkuan Song
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace frozen similarity retriever with jointly trained Mixture-of-Logits retrieval (ablation) | JRPP · ICIP, SMPD, Instagram | MSE/MAE/SRC | Improved |
| add retrieval module (ablation) | JRPP · ICIP, SMPD, Instagram | MSE/MAE/SRC | Improved |
| add uncertainty-aware filter (Gumbel-Softmax selection plus IB refinement) (ablation) | JRPP · ICIP, SMPD, Instagram | MSE/MAE/SRC | Improved |
| add confidence-weighted test-time perturbation (ablation) | JRPP · ICIP, SMPD, Instagram | MSE/MAE/SRC | Improved |
| use JRPP instead of SKAPP | JRPP · Instagram | MSE: 2.0936 -> 1.5627 | Improved |
| use JRPP instead of SKAPP | JRPP · ICIP | SRC: 0.6965 -> 0.7125 | Improved |
| use JRPP instead of SKAPP on tail UGCs (top 10%) | JRPP · ICIP (top 10% most popular) | MSE: 4.907 -> 3.712 | Improved |
| increase number of retrieved UGCs k beyond moderate value (ablation) | JRPP · ICIP | MSE | Hurt |
| increase number of low-rank embeddings M to large values (ablation) | JRPP · ICIP | MSE | Hurt |
| increase number of test-time perturbations T (ablation) | JRPP · ICIP | MSE | No clear effect |
| increase number of low-rank embeddings M from small to moderate (ablation) | JRPP · ICIP | MSE | Improved |
Closest papers by meaning
- Improving Multimodal Social Media Popularity Prediction via Selective Retrieval Knowledge Augmentation (AAAI 2025 · similarity 0.93)
- THGFormer: Time-Aware Hypergraph Learning for Multimodal Social Media Popularity Prediction (Student Abstract) (AAAI 2024 · similarity 0.86)
- SMTPD: A New Benchmark for Temporal Prediction of Social Media Popularity (CVPR 2025 · similarity 0.86)
- In-context Prompt-augmented Micro-video Popularity Prediction (AAAI 2025 · similarity 0.86)
- Drift-Aware Multimodal User Representation Learning via Multi-Scale Temporal Modeling and Sparse Mixture-of-Experts (NeurIPS 2026 · similarity 0.84)
- Influence Guided Context Selection for Effective Retrieval-Augmented Generation (NeurIPS 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
