Research map / Architectures and efficiency / Attention and transformers
Keywords-Guided Abstractive Sentence 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 |
|---|---|---|---|
| add keywords as input via merging into the sentence (ablation) | S2S-Sentence · Gigaword test | ROUGE-2: 23.58 -> 24.07 | Helped |
| use keywords only as input instead of sentence (ablation) | S2S-Sentence · Gigaword test | ROUGE-2: 23.58 -> 20.54 | Hurt |
| use keywords-guided co-selective encoding instead of self-selective (ablation) | Hier fusion · Gigaword test | ROUGE-2: 24.32 -> 24.75 | Helped |
| use hierarchical fusion instead of concatenation (ablation) | Co-Selective+DualPG · Gigaword test | ROUGE-2: 24.39 -> 25.06 | Helped |
| add dual-copy pointer instead of single-copy PG (ablation) | Co-Selective+Hier · Gigaword test | ROUGE-2: 24.83 -> 25.06 | Helped |
| add pointer-generator copy (ablation) | Co-Selective+Hier · Gigaword test | ROUGE-2: 24.75 -> 24.83 | Helped |
| train with multi-task learning instead of two-stage learning (ablation) | Co-Selective models · Keyword extraction | F1: 59.35 -> 60.27 | Helped |
| MTL instead of two-stage learning for summarization (ablation) | Co-Selective models · Gigaword test | ROUGE-2 | Helped |
| fine-tune with predicted keywords (ablation) | Co-Selective Hier · Gigaword test | ROUGE-2: about 1% absolute | Helped |
| replace BiLSTM-softmax with BiLSTM-CRF keyword extractor (ablation) | keyword extractor · Keyword extraction | accuracy: +0.2% | No clear effect |
| use extractor keywords instead of TextRank for KIGN (ablation) | KIGN · Gigaword test | ROUGE-1: 46.18 -> 46.84 | Helped |
| use ground-truth keywords at train and test (oracle) (ablation) | Hier+DualPG · Gigaword test | ROUGE-2: 25.06 -> 42.94 | Helped |
Closest papers by meaning
- Joint Parsing and Generation for Abstractive Summarization (AAAI 2020 · similarity 0.90)
- Unity in Diversity: Learning Distributed Heterogeneous Sentence Representation for Extractive Summarization (AAAI 2018 · similarity 0.88)
- A Deep Reinforced Model for Abstractive Summarization (ICLR 2018 · similarity 0.88)
- Summarization Programs: Interpretable Abstractive Summarization with Neural Modular Trees (ICLR 2023 · similarity 0.88)
- Towards Improving Faithfulness in Abstractive Summarization (NeurIPS 2022 · similarity 0.88)
- Distilling Structured Rationale from Large Language Models to Small Language Models for Abstractive Summarization (AAAI 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.