Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Text Summarization with Oracle Expectation
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 expectation-based Oreo sentence labels instead of greedy labels | BertSum · CNN/DM, XSum, Multi-News, Reddit, WikiHow test sets | ROUGE-L: 0.18--0.89 Rouge-L improvement | Helped |
| Use beam labels instead of greedy labels | BertSum · WikiHow test | ROUGE-L: 28.24 -> 25.71 | Hurt |
| Use Oreo labels instead of greedy labels | BertSum trained on CNN/DM · XSum zero-shot | ROUGE-L: 15.62 -> 15.58 | No clear effect |
| Use Oreo labels instead of greedy labels | BertSum trained on CNN/DM · Reddit zero-shot | ROUGE-L: 17.06 -> 17.71 | Helped |
| Use Oreo labels instead of greedy labels | BertSum trained on CNN/DM · WikiHow zero-shot | ROUGE-L: 25.39 -> 25.62 | Helped |
| Use Oreo labels instead of greedy labels | XlS · MLSum German and Turkish zero-shot | ROUGE-L: 2.72 on German; 2.19 on Turkish | Helped |
| Use Oreo guidance instead of greedy guidance | GSum · CNN/DM test | ROUGE-L: 0.37 Rouge-L points | Helped |
| Replace expectation aggregation with max aggregation (Ormax) (ablation) | BertSum + Oreo · CNN/DM test | ROUGE-L: ↓0.20 | Hurt |
| Replace expectation aggregation with max aggregation (Ormax) (ablation) | GSum + Oreo · CNN/DM test | ROUGE-L: ↓0.00 | No clear effect |
| Use label smoothing with greedy hard labels (ablation) | BertSum · CNN/DM test | ROUGE-L: ↑0.03 | No clear effect |
| Use label smoothing with beam hard labels (ablation) | BertSum · CNN/DM test | ROUGE-L: ↑0.01 | No clear effect |
| Use uniform oracle weights rather than hand-crafted rank, quality, locality, or position schedules (ablation) | Oreo · CNN/DM validation | ROUGE-L: 40.69; alternatives: 40.50, 40.60, 40.42, 40.26 | Helped |
Closest papers by meaning
- SummaRuNNer: A Recurrent Neural Network Based Sequence Model for Extractive Summarization of Documents (AAAI 2017 · similarity 0.88)
- Unsupervised Extractive Summarization with Learnable Length Control Strategies (AAAI 2024 · similarity 0.87)
- Document Summarization with Conformal Importance Guarantees (NeurIPS 2025 · similarity 0.84)
- Improving Faithfulness in Abstractive Text Summarization with EDUs Using BART (Student Abstract) (AAAI 2024 · similarity 0.84)
- Language Models of Code Are Few-Shot Planners and Reasoners for Multi-Document Summarization with Attribution (AAAI 2025 · similarity 0.84)
- BooookScore: A systematic exploration of book-length summarization in the era of LLMs (ICLR 2024 · 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.