Research map / Label-efficient and robust learning / Knowledge distillation
Flexible Non-Autoregressive Extractive Summarization with Threshold: How to Extract a Non-Fixed Number of Summary Sentences
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 |
|---|---|---|---|
| use ThresSum's threshold-based extraction with iterative refinement and soft-label distillation instead of MATCHSUM | CNN/DM extractive summarization · CNN/DM | ROUGE-1: 44.41 -> 44.59 | Helped |
| use ThresSum's threshold-based extraction with iterative refinement and soft-label distillation instead of MATCHSUM | CNN/DM extractive summarization · CNN/DM | ROUGE-2: 20.86 -> 21.15 | Helped |
| use ThresSum's threshold-based extraction with iterative refinement and soft-label distillation instead of MATCHSUM | CNN/DM extractive summarization · CNN/DM | ROUGE-L: 40.55 -> 40.76 | Helped |
| use ThresSum instead of BERTSUMEXT | NYT extractive summarization · NYT | ROUGE-1: 48.51 -> 50.08 | Helped |
| use ThresSum instead of BERTSUMEXT | NYT extractive summarization · NYT | ROUGE-2: 30.27 -> 31.77 | Helped |
| use ThresSum instead of BERTSUMEXT | NYT extractive summarization · NYT | ROUGE-L: 44.65 -> 45.21 | Helped |
| add knowledge distillation with progressive soft labels (ablation) | ThresSum · CNN/DM | ROUGE-1: 44.18 -> 44.59 | Helped |
| add knowledge distillation with progressive soft labels (ablation) | ThresSum · CNN/DM | ROUGE-2: 20.95 -> 21.15 | Helped |
| add knowledge distillation with progressive soft labels (ablation) | ThresSum · CNN/DM | ROUGE-L: 40.42 -> 40.76 | Helped |
| use ALBERT instead of BERT (ablation) | ThresSum · CNN/DM | ROUGE-1: 44.35 -> 44.59 | Helped |
| use ALBERT instead of BERT (ablation) | ThresSum · CNN/DM | ROUGE-2: 21.03 -> 21.15 | Helped |
| use ALBERT instead of BERT (ablation) | ThresSum · CNN/DM | ROUGE-L: 40.57 -> 40.76 | Helped |
Closest papers by meaning
- Unsupervised Extractive Summarization with Learnable Length Control Strategies (AAAI 2024 · similarity 0.88)
- SummaRuNNer: A Recurrent Neural Network Based Sequence Model for Extractive Summarization of Documents (AAAI 2017 · similarity 0.87)
- Unity in Diversity: Learning Distributed Heterogeneous Sentence Representation for Extractive Summarization (AAAI 2018 · similarity 0.87)
- Distilling Structured Rationale from Large Language Models to Small Language Models for Abstractive Summarization (AAAI 2025 · similarity 0.84)
- RECOMP: Improving Retrieval-Augmented LMs with Context Compression and Selective Augmentation (ICLR 2024 · similarity 0.84)
- Document Summarization with Conformal Importance Guarantees (NeurIPS 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.