Research map / Architectures and efficiency / Attention and transformers
Translation Prediction with Source Dependency-Based Context Representation
Kehai Chen, Tiejun Zhao, Muyun Yang, Lemao Liu
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 DBiCSNNLM to baseline SMT | Moses baseline · NIST Chinese-English average (MT03/04/05) | BLEU: 34.37 -> 36.28 | Improved |
| replace NNJM with DBiCSNNLM | SMT with neural model · NIST Chinese-English average (MT03/04/05) | BLEU: 35.60 -> 36.28 | Improved |
| use BiCNN representation instead of feedforward over DBiCU tokens (DBiCSFFNN) (ablation) | DBiCSNNLM · NIST Chinese-English average (MT03/04/05) | BLEU: 35.36 -> 36.28 | Improved |
| use neural DBiCSNNLM instead of 5-gram DBiCSLM (ablation) | dependency-based LM · NIST Chinese-English average (MT03/04/05) | BLEU: 35.28 -> 36.28 | Improved |
| add source dependency context (DBiCSLM vs BiLM) | bilingual LM · NIST Chinese-English average (MT03/04/05) | BLEU: 34.80 -> 35.28 | Improved |
| add DBiCSNNLM on top of NNLM baseline | baseline+NNLM · NIST average with NNLM baseline | BLEU: 34.97 -> 36.64 | Improved |
| add NNLM to baseline+NNJM (ablation) | NNJM · NIST average | BLEU | No clear effect |
| DBiCSNNLM rescoring vs MTURNN | 1000-best rescoring · NIST average, 1000-best rescoring | BLEU: 35.05 -> 35.12 | Improved |
Closest papers by meaning
- Improving Context-Aware Neural Machine Translation Using Self-Attentive Sentence Embedding (AAAI 2020 · similarity 0.86)
- Syntax-Directed Attention for Neural Machine Translation (AAAI 2018 · similarity 0.86)
- Towards Reliable Neural Machine Translation with Consistency-Aware Meta-Learning (AAAI 2023 · similarity 0.84)
- Exploring the Translation Mechanism of Large Language Models (NeurIPS 2025 · similarity 0.84)
- Quantifying the Plausibility of Context Reliance in Neural Machine Translation (ICLR 2024 · similarity 0.83)
- A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models (ICLR 2024 · 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.
