FORTAP: Using Formulas for Numerical-Reasoning-Aware Table Pretraining

TL;DR

FORTAP introduces a novel pretraining approach for tables that leverages spreadsheet formulas to enhance numerical reasoning capabilities, achieving state-of-the-art results on key tasks.

Contribution

It is the first to utilize large-scale spreadsheet formulas for numerical reasoning-aware table pretraining, guiding models to learn calculations and references.

Findings

Achieves state-of-the-art on cell type classification

Improves formula prediction accuracy

Demonstrates potential of formula-based pretraining

Abstract

Tables store rich numerical data, but numerical reasoning over tables is still a challenge. In this paper, we find that the spreadsheet formula, which performs calculations on numerical values in tables, is naturally a strong supervision of numerical reasoning. More importantly, large amounts of spreadsheets with expert-made formulae are available on the web and can be obtained easily. FORTAP is the first method for numerical-reasoning-aware table pretraining by leveraging large corpus of spreadsheet formulae. We design two formula pretraining tasks to explicitly guide FORTAP to learn numerical reference and calculation in semi-structured tables. FORTAP achieves state-of-the-art results on two representative downstream tasks, cell type classification and formula prediction, showing great potential of numerical-reasoning-aware pretraining.