Exploring Generative Process Reward Modeling for Semi-Structured Data: A Case Study of Table Question Answering

TL;DR

This paper investigates the use of process reward models for semi-structured data, specifically table question answering, revealing their potential and current limitations in this domain.

Contribution

First systematic study of process reward models applied to table question answering, analyzing their effectiveness and challenges in semi-structured data tasks.

Findings

PRMs with textual and code verification aid answer selection

PRMs struggle to generalize to out-of-domain data

Weak correlation between step verification and answer accuracy

Abstract

Process reward models (PRMs) enhance complex reasoning in large language models (LLMs) by evaluating candidate solutions step-by-step and selecting answers based on aggregated step scores. While effective in domains such as mathematics, their applicability to tasks involving semi-structured data, like table question answering (TQA), remains unexplored. TQA poses unique challenges for PRMs, including abundant irrelevant information, loosely connected reasoning steps, and domain-specific reasoning. This work presents the first systematic study of PRMs for TQA. We evaluate state-of-the-art generative PRMs on TQA from both answer and step perspectives. Results show that PRMs that combine textual and code verification can aid solution selection but struggle to generalize to out-of-domain data. Analysis reveals a weak correlation between performance in step-level verification and answer accuracy, possibly stemming from weak step dependencies and loose causal links. Our findings highlight limitations of current PRMs on TQA and offer valuable insights for building more robust, process-aware verifiers.