This preprint presents a structure-aware system for automatic summarization and question answering over scientific research papers. The system leverages transformer-based models, including BART-large-CNN for abstractive summarization and RoBERTa for extractive question answering. It identifies standard research paper sections, applies section-wise summarization with overlapping text chunking, and integrates intelligent context selection for accurate question answering. The implementation emphasizes robustness through fallback mechanisms, conservative parameter tuning, and reliable PDF text and metadata extraction using PyMuPDF. Designed for real-world usability, the system enables researchers to efficiently extract key insights and query complex documents using natural language.
Saha et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: