Meow: End-to-End Outline Writing for Automatic Academic Survey
TLDR
Meow is a metadata-driven framework for end-to-end outline writing in automatic academic survey generation, using a curated dataset and two-stage training.
Reasoning
The paper introduces a novel end-to-end approach to outline writing, addressing limitations of template-based methods with a metadata-driven framework, curated dataset, and reinforcement learning. Strengths include clear problem formulation and systematic evaluation, but the scope is limited to outline writing rather than full survey generation, and real-world validation is based on dataset curation rather than deployed experiments.
Read-first score
Read-first score 50.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 46.
Field roles
Rank sensitivity
Stability: volatile; rank range: 27.
Keyword Scores
Deep Analysis
Innovations
- First metadata-driven outline writing framework for automatic survey generation
- Formulation of outline writing as an end-to-end task generating hierarchical structured outlines from paper metadata
- Curation of a high-quality dataset from arXiv, bioRxiv, and medRxiv
- Establishment of systematic evaluation metrics for outline quality assessment
- Two-stage training approach combining supervised fine-tuning and reinforcement learning
Methodology
Meow formulates outline writing as an end-to-end task that generates hierarchical structured outlines from paper metadata. A dataset is curated from arXiv, bioRxiv, and medRxiv, and systematic evaluation metrics are established. An 8B reasoning model is trained using a two-stage approach: supervised fine-tuning followed by reinforcement learning.
Key Results
The 8B reasoning model achieves strong performance with high structural fidelity and stylistic coherence.