Improving Attributed Long-form Question Answering with Intent Awareness

TL;DR

This paper introduces intent awareness in large language models to improve the quality of long-form scientific report generation, resulting in better performance, citation accuracy, and readability.

Contribution

It develops structured schemes to extract author intents, enhancing zero-shot generation and synthetic data creation for smaller models, advancing scientific report synthesis.

Findings

Improved performance by +2.9 and +12.3 points for large and small models.

Enhanced citation usage and report readability.

Effective intent extraction boosts model capabilities.

Abstract

Large language models (LLMs) are increasingly being used to generate comprehensive, knowledge-intensive reports. However, while these models are trained on diverse academic papers and reports, they are not exposed to the reasoning processes and intents that guide authors in crafting these documents. We hypothesize that enhancing a model's intent awareness can significantly improve the quality of generated long-form reports. We develop and employ structured, tag-based schemes to better elicit underlying implicit intents to write or cite. We demonstrate that these extracted intents enhance both zero-shot generation capabilities in LLMs and enable the creation of high-quality synthetic data for fine-tuning smaller models. Our experiments reveal improved performance across various challenging scientific report generation tasks, with an average improvement of +2.9 and +12.3 absolute points for large and small models over baselines, respectively. Furthermore, our analysis illuminates how intent awareness enhances model citation usage and substantially improves report readability.