Towards Probabilistic Question Answering Over Tabular Data

TL;DR

This paper introduces a new benchmark and framework for probabilistic question answering over large tabular data, combining Bayesian Networks and large language models to improve reasoning under uncertainty.

Contribution

It presents a novel approach that induces Bayesian Networks from tables and translates natural language queries into probabilistic queries, advancing probabilistic QA methods.

Findings

Significant performance improvements over baselines

Effective hybrid symbolic-neural reasoning approach

Demonstrated benefits of probabilistic reasoning in QA

Abstract

Current approaches for question answering (QA) over tabular data, such as NL2SQL systems, perform well for factual questions where answers are directly retrieved from tables. However, they fall short on probabilistic questions requiring reasoning under uncertainty. In this paper, we introduce a new benchmark LUCARIO and a framework for probabilistic QA over large tabular data. Our method induces Bayesian Networks from tables, translates natural language queries into probabilistic queries, and uses large language models (LLMs) to generate final answers. Empirical results demonstrate significant improvements over baselines, highlighting the benefits of hybrid symbolic-neural reasoning.