Planning for Success: Exploring LLM Long-term Planning Capabilities in Table Understanding

TL;DR

This paper introduces a method leveraging LLMs' long-term planning capabilities to improve table understanding, addressing limitations of prior Chain-of-Thought approaches, and achieves state-of-the-art results on key datasets.

Contribution

It proposes a novel long-term planning approach for LLMs in table understanding, enhancing inter-step connections and goal alignment over previous methods.

Findings

Outperforms strong baselines on WikiTableQuestions

Achieves state-of-the-art on TabFact dataset

Effectively minimizes unnecessary details in reasoning steps

Abstract

Table understanding is key to addressing challenging downstream tasks such as table-based question answering and fact verification. Recent works have focused on leveraging Chain-of-Thought and question decomposition to solve complex questions requiring multiple operations on tables. However, these methods often suffer from a lack of explicit long-term planning and weak inter-step connections, leading to miss constraints within questions. In this paper, we propose leveraging the long-term planning capabilities of large language models (LLMs) to enhance table understanding. Our approach enables the execution of a long-term plan, where the steps are tightly interconnected and serve the ultimate goal, an aspect that methods based on Chain-of-Thought and question decomposition lack. In addition, our method effectively minimizes the inclusion of unnecessary details in the process of solving the next short-term goals, a limitation of methods based on Chain-of-Thought. Extensive experiments demonstrate that our method outperforms strong baselines and achieves state-of-the-art performance on WikiTableQuestions and TabFact datasets.