Can LLMs Narrate Tabular Data? An Evaluation Framework for Natural Language Representations of Text-to-SQL System Outputs

Can LLMs Narrate Tabular Data? An Evaluation Framework for Natural Language Representations of Text-to-SQL System Outputs

Jinghwi Singh, Wei Sun, Amit Agarwal, Venkat Krishnamurthy, Yassine Benajiba, Sujith Ravi, Dan Roth. EMNLP Findings (2025).

Abstract. This paper evaluates whether large language models can faithfully turn Text-to-SQL outputs into natural-language descriptions. It proposes a framework for judging the quality of such narrations over tabular results.

Updated: