New framework uses discourse structure for better long document QA

Researchers have developed a novel discourse-aware hierarchical framework for question answering on long documents. This approach utilizes Rhetorical Structure Theory (RST) to parse discourse trees and enhance sentence-level representations with LLM-based node embeddings. The framework's structure-guided hierarchical retrieval method has demonstrated consistent improvements across various datasets, genres, and languages, outperforming existing methods by incorporating discourse structure. AI

IMPACT Introduces a novel method for long document QA that leverages discourse structure, potentially improving comprehension and retrieval accuracy for complex texts.

RANK_REASON This is a research paper detailing a new framework for long document question answering. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• LLM
• Rhetorical Structure Theory

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →