HRoT: Hybrid prompt strategy and Retrieval of Thought for Table-Text Hybrid Question Answering

TL;DR

This paper introduces a novel hybrid prompt strategy with retrieval of thought to enhance large language models' ability to answer numerical questions over hybrid table and text data, achieving superior few-shot performance.

Contribution

It proposes a new prompting method combining hybrid prompts and retrieval of thought, improving LLMs' reasoning over hybrid data in TextTableQA tasks.

Findings

Outperforms fully-supervised SOTA on MultiHiertt dataset in few-shot setting

Enhances retrieval thinking ability in LLMs for hybrid data

Demonstrates effectiveness of hybrid prompts in TextTableQA

Abstract

Answering numerical questions over hybrid contents from the given tables and text(TextTableQA) is a challenging task. Recently, Large Language Models (LLMs) have gained significant attention in the NLP community. With the emergence of large language models, In-Context Learning and Chain-of-Thought prompting have become two particularly popular research topics in this field. In this paper, we introduce a new prompting strategy called Hybrid prompt strategy and Retrieval of Thought for TextTableQA. Through In-Context Learning, we prompt the model to develop the ability of retrieval thinking when dealing with hybrid data. Our method achieves superior performance compared to the fully-supervised SOTA on the MultiHiertt dataset in the few-shot setting.