Research map / Large language models / Language, safety and interpretability
Controlling the Amount of Verbatim Copying in Abstractive Summarization
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 |
|---|---|---|---|
| Train with more unseen summary tokens relative to seen tokens (ablation) | CopyTrans · Gigaword and Newsroom | Average n-gram copy rate: 52.25 -> 39.44 (Gigaword); 60.22 -> 48.87 (Newsroom) | Hurt |
| Train with more unseen summary tokens relative to seen tokens (ablation) | CopyTrans · Gigaword and Newsroom | ROUGE-2: 14.51 -> 20.00 (Gigaword); 21.51 -> 23.90 (Newsroom) | Helped |
| Add source-token prediction to training (ablation) | CopyTrans · Gigaword and Newsroom | ROUGE-2: 15.21 -> 19.55 (Gigaword); 21.81 -> 24.14 (Newsroom) | Helped |
| Use beam search instead of best-first search (ablation) | CopyTrans · Gigaword test | ROUGE-2: 20.28 -> 20.37 | No clear effect |
| Apply BP-norm reranking instead of unreranked beam search (ablation) | CopyTrans · Gigaword test | ROUGE-1: 38.87 -> 39.19 | Helped |
| Apply SBWR reranking instead of unreranked beam search (ablation) | CopyTrans · Gigaword test | ROUGE-2: 20.37 -> 20.47 | Helped |
| Train to favor copying rather than a more abstractive output (ablation) | CopyTrans · Newsroom test, Gigaword-trained | ROUGE-1: 40.89 -> 39.44 | Hurt |
Closest papers by meaning
- Faithful to the Original: Fact Aware Neural Abstractive Summarization (AAAI 2018 · similarity 0.90)
- Joint Parsing and Generation for Abstractive Summarization (AAAI 2020 · similarity 0.90)
- Exploring Explainable Selection to Control Abstractive Summarization (AAAI 2021 · similarity 0.90)
- Summarization Programs: Interpretable Abstractive Summarization with Neural Modular Trees (ICLR 2023 · similarity 0.89)
- Towards Improving Faithfulness in Abstractive Summarization (NeurIPS 2022 · similarity 0.89)
- Learning to Summarize by Learning to Quiz: Adversarial Agentic Collaboration for Long Document Summarization (ICLR 2026 · similarity 0.86)
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.