AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Semi-Supervised Synthetic Data Generation with Fine-Grained Relevance Control for Short Video Search Relevance Modeling

Haoran Li, Zhiming Su, Junyan Yao, Enwei Zhang, Yang Ji, Yan Chen, Kan Zhou, Chao Feng, Jiao Ran

AAAI 2026 · Label-efficient and robust learning › Semi-supervised learning · approach: Self-training theory · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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 changedWhereMeasuredEffect
add SSRA synthetic data to annotated training dataQwen3-Embedding-4B · Short video retrievalnDCG@10: 73.20 -> 74.47Improved
add prompt-based SyCL Modified synthetic dataQwen3-Embedding-4B · Short video pair classificationAP average: 69.57 -> 67.95Hurt
add vanilla SFT query-model synthetic dataQwen3-Embedding-4B · Short video pair classificationAP average: 69.57 -> 70.87Improved
add SSRA over vanilla SFT synthesisQwen3-Embedding-4B · Short video pair classificationAP average: 70.87 -> 71.52Improved
add SSRA stage 1 (score-based re-annotation) (ablation)SSRA · Short video retrievalnDCG@10: 74.02 -> 74.23 (4B)Improved
add SSRA stage 2 (relevance alignment filtering) (ablation)SSRA · Short video pair classificationAP: 71.30 -> 74.92 (4B)Improved
add SSRA stage 1 only (without stage 2) for pair classification (ablation)Qwen3-Embedding-0.6B · Short video pair classificationAP: 70.70 -> 70.77No clear effect
use stage 1 re-labeled data for query model (ablation)query model · 10,000 sampled documentsduplicate query rate: 6.57% -> 5.20%Improved
add stage 2 training to query model (ablation)query model · 200 documents, human annotatedqueries matching intended label (label 1): 81 / 200 -> 130 / 200Improved
add stage 2 training to query model (ablation)query model · 200 documents, human annotatedqueries matching intended label (label 3): 189 / 200 -> 178 / 200Hurt
include intermediate relevance labels 1 and 2 in synthetic data (multi-level vs binary) (ablation)Qwen3-Embedding-4B · Short video retrievalnDCG@10: 73.73 -> 74.23Improved
include intermediate relevance labels 1 and 2 in synthetic data (multi-level vs binary) (ablation)Qwen3-Embedding-4B · Balanced pair classificationAP@>=3: 58.39 -> 55.25Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy