Research map / Architectures and efficiency / Attention and transformers
Generating Live Soccer-Match Commentary from Play Data
Yasufumi Taniguchi, Yukun Feng, Hiroya Takamura, Manabu Okumura
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add placeholders for player/team names (Model 2 vs 1) (ablation) | encoder-decoder baseline · Soccer live commentary test set | BLEU: 0.11 -> 0.43 | Improved |
| add placeholders for player/team names (Model 2 vs 1) (ablation) | encoder-decoder baseline · Soccer live commentary test set (human eval) | informativeness: 1.41 -> 1.38 | No clear effect |
| add gate mechanism to placeholder model (Model 3 vs 2) (ablation) | Model 2 · Soccer live commentary test set | BLEU: 0.43 -> 0.58 | Improved |
| add gate mechanism to placeholder+reconstruction model (Model 5 vs 4) (ablation) | Model 4 · Soccer live commentary test set | BLEU (l<=10): 6.38 -> 7.46 | Improved |
| add gate mechanism to placeholder+reconstruction model (Model 5 vs 4) (ablation) | Model 4 · Soccer live commentary test set | BLEU: 0.65 -> 0.51 | Hurt |
| add reconstruction error to placeholder model (Model 4 vs 2) (ablation) | Model 2 · Soccer live commentary test set | BLEU: 0.43 -> 0.65 | Improved |
| add reconstruction error to gate+placeholder model (Model 5 vs 3) (ablation) | Model 3 · Soccer live commentary test set | BLEU (l<=10): 5.19 -> 7.46 | Improved |
| add reconstruction error to gate+placeholder+word concat (Model 7 vs 6) (ablation) | Model 6 · Soccer live commentary test set | BLEU: 0.12 -> 0.69 | Improved |
| add word concatenation of frequent bigrams (Model 7 vs 5) (ablation) | Model 5 · Soccer live commentary test set | BLEU: 0.51 -> 0.69 | Improved |
| add word concatenation of frequent bigrams (Model 7 vs 5) (ablation) | Model 5 · Soccer live commentary test set | BLEU (l<=10): 7.46 -> 6.79 | Hurt |
| add reconstruction error (Model 5 vs 3) (ablation) | Model 3 · Soccer live commentary test set (human eval) | grammaticality: 1.89 -> 1.99 | Improved |
| full model vs baseline (Model 7 vs 1) (ablation) | baseline · Soccer live commentary test set (human eval) | informativeness: 1.41 -> 1.69 | Improved |
Closest papers by meaning
- Towards Universal Soccer Video Understanding (CVPR 2025 · similarity 0.88)
- Event2Tracking: Reconstructing Multi-Agent Soccer Trajectories Using Long-Term Multimodal Context (AAAI 2025 · similarity 0.86)
- SoccerMaster: A Vision Foundation Model for Soccer Understanding (CVPR 2026 · similarity 0.84)
- SportR: A Benchmark for Multimodal Large Language Model Reasoning in Sports (ICLR 2026 · similarity 0.83)
- SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models (ICLR 2025 · similarity 0.82)
- Sports-Traj: A Unified Trajectory Generation Model for Multi-Agent Movement in Sports (ICLR 2025 · similarity 0.82)
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.
