Research map / Label-efficient and robust learning / Recognition and long-tail
Cross-Lingual Universal Dependency Parsing Only From One Monolingual Treebank
Kailai Sun, Zuchao Li, Hai Zhao
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 WO and MLM objectives (multi-task) (ablation) | biaffine parser · PTB | UAS: 95.83 -> 96.05 | Improved |
| add WO and MLM objectives (multi-task) (ablation) | biaffine parser · CTB | UAS: 90.32 -> 91.22 | Improved |
| add WO objective (ablation) | Ours · CTB | UAS: 91.04 -> 91.22 | Improved |
| add MLM objective (ablation) | Ours · PTB | UAS: 95.92 -> 96.05 | Improved |
| multi-task training instead of pipeline pretraining (ablation) | Ours · CTB | UAS: 90.70 -> 91.22 | Improved |
| add WO and MLM objectives (ablation) | Ours · en->de zero-shot first round | UAS: 70.23 -> 71.16 | Improved |
| add source-language modeling (ablation) | Ours · en->de final round | UAS: 75.74 -> 76.20 | Improved |
| add source-language modeling (ablation) | Ours · en->fr final round | UAS: 82.83 -> 83.15 | Improved |
| self-training with multilingual BERT | Ours · UD 21 target languages (avg) | UAS: 70.25 -> 80.32 | Improved |
| Ours versus Ahmad et al. cross-lingual | Ours · UD 21 target languages (avg) | UAS: 63.22 -> 70.25 | Improved |
| add mBERT to monolingual parser | Ours · UD monolingual de/no/nl/ru | UAS: 90.26 vs 89.48 | Mixed |
| add multi-task LM objectives | BIAF · UD en->Spoken fr (cross-domain) | UAS: 72.54 -> 77.73 | Improved |
Closest papers by meaning
- A Representation Learning Framework for Multi-Source Transfer Parsing (AAAI 2016 · similarity 0.90)
- Dynamic Syntactic Feature Filtering and Injecting Networks for Cross-lingual Dependency Parsing (AAAI 2025 · similarity 0.89)
- Multilingual Pre-training with Universal Dependency Learning (NeurIPS 2021 · similarity 0.88)
- Global Greedy Dependency Parsing (AAAI 2020 · similarity 0.87)
- Bidirectional Transition-Based Dependency Parsing (AAAI 2019 · similarity 0.87)
- Implicit Word Reordering with Knowledge Distillation for Cross-Lingual Dependency Parsing (AAAI 2025 · similarity 0.84)
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.
