Research map / Graph learning / Knowledge and temporal graphs
Structure Learning for Headline Generation
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 interaction-focused instead of representation-focused matching (ablation) | SLGen (ReLU) · NYT Annotated corpus test | ROUGE-1: 34.42 -> 35.82 | Helped |
| use ReLU activation instead of k-max pooling for edges (ablation) | SLGen (interaction) · NYT Annotated corpus test | ROUGE-1: 35.46 -> 35.82 | Helped |
| add deep (semantic matching) component (ablation) | SLGenInt+Relu · NYT Annotated corpus test | ROUGE-1: 34.27 -> 35.82 | Helped |
| add wide (discourse) component (ablation) | SLGenInt+Relu · NYT Annotated corpus test | ROUGE-1: 35.77 -> 35.82 | No clear effect |
| learn sentence graph instead of tf-idf similarity graph (ablation) | SLGen · NYT Annotated corpus test | ROUGE-1: 34.30 -> 35.82 | Helped |
| learn sentence graph instead of ADG discourse graph (ablation) | SLGen · NYT Annotated corpus test | ROUGE-1: 34.31 -> 35.82 | Helped |
| mix tf-idf graph and ADG (ablation) | SLGen · NYT Annotated corpus test | ROUGE-1: 34.35 vs 34.30 (tfidf alone) | No clear effect |
| SLGenInt+Relu vs BILSTM+GNN (pre-designed graph) | headline generation · NYT Annotated corpus test | ROUGE-1: 34.11 -> 35.82 | Helped |
| SLGenInt+Relu vs BILSTM+GNN | headline generation · NYT Annotated corpus test | ROUGE-L: 32.89 -> 34.12 | Helped |
Closest papers by meaning
- Structured Neural Summarization (ICLR 2019 · similarity 0.87)
- Graph-KV: Breaking Sequence via Injecting Structural Biases into Large Language Models (NeurIPS 2025 · similarity 0.85)
- SSTAG: Structure-Aware Self-Supervised Learning Method for Text-Attributed Graphs (NeurIPS 2025 · similarity 0.85)
- Attention Mechanisms Perspective: Exploring LLM Processing of Graph-Structured Data (ICML 2025 · similarity 0.84)
- Scale-Free Graph-Language Models (ICLR 2025 · similarity 0.84)
- GOFA: A Generative One-For-All Model for Joint Graph Language Modeling (ICLR 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.