Discourse-Aware Unsupervised Summarization of Long Scientific Documents
This work addresses the challenge of summarizing scientific articles for researchers and readers, offering an unsupervised approach that reduces the need for large labeled datasets.
The authors tackled the problem of summarizing long scientific documents without supervision by proposing a graph-based ranking model that uses discourse structure and positional cues, achieving performance comparable to supervised methods on PubMed and arXiv datasets.
We propose an unsupervised graph-based ranking model for extractive summarization of long scientific documents. Our method assumes a two-level hierarchical graph representation of the source document, and exploits asymmetrical positional cues to determine sentence importance. Results on the PubMed and arXiv datasets show that our approach outperforms strong unsupervised baselines by wide margins in automatic metrics and human evaluation. In addition, it achieves performance comparable to many state-of-the-art supervised approaches which are trained on hundreds of thousands of examples. These results suggest that patterns in the discourse structure are a strong signal for determining importance in scientific articles.