Research map / Architectures and efficiency / Attention and transformers
A Task in a Suit and a Tie: Paraphrase Generation with Semantic Augmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 frame and role semantic encoders | Transformer paraphrase generation · MSCOCO | BLEU: 41.0 -> 44.0 | Helped |
| Add frame and role semantic encoders | Transformer paraphrase generation · WikiAnswers | BLEU: 41.9 -> 43.9 | Helped |
| Use frame annotations alone (ablation) | Transformer paraphrase generation · MSCOCO | BLEU: 41.0 -> 42.5 | Helped |
| Use role annotations alone (ablation) | Transformer paraphrase generation · MSCOCO | BLEU: 41.0 -> 41.8 | Helped |
| Combine frame and role annotations (ablation) | Transformer paraphrase generation · MSCOCO | BLEU: 41.0 -> 44.0 | Helped |
| Use frame annotations alone (ablation) | Transformer paraphrase generation · WikiAnswers | BLEU: 41.9 -> 43.2 | Helped |
| Use role annotations alone (ablation) | Transformer paraphrase generation · WikiAnswers | BLEU: 41.9 -> 43.0 | Helped |
| Combine frame and role annotations (ablation) | Transformer paraphrase generation · WikiAnswers | BLEU: 41.9 -> 43.9 | Helped |
| Add semantic augmentation | Transformer paraphrase generation · Sentences >20 words | BLEU: 17.7 -> 17.9; 16.8 -> 16.4 | No clear effect |
| Use Transformer instead of NV-LSTM | Paraphrase generation · MSCOCO | BLEU: 41.1 -> 41.0 | No clear effect |
| Use Transformer instead of NV-LSTM | Paraphrase generation · MSCOCO human acceptability, target and prediction | acceptability: 19.0 -> 31.1 | Helped |
| Add semantic augmentation | Transformer paraphrase generation · MSCOCO human evaluation, target, source, and prediction | acceptability: 45.6 -> 66.5 | Helped |
Closest papers by meaning
- Integrating Linguistic Knowledge to Sentence Paraphrase Generation (AAAI 2020 · similarity 0.87)
- A Deep Generative Framework for Paraphrase Generation (AAAI 2018 · similarity 0.87)
- Latent Paraphrasing: Perturbation on Layers Improves Knowledge Injection in Language Models (NeurIPS 2024 · similarity 0.85)
- A Lens into Interpretable Transformer Mistakes via Semantic Dependency (ICML 2025 · similarity 0.84)
- LAMPAT: Low-Rank Adaption for Multilingual Paraphrasing Using Adversarial Training (AAAI 2024 · similarity 0.84)
- Adversarial Paraphrasing: A Universal Attack for Humanizing AI-Generated Text (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.