FORTAP: Using Formulas for Numerical-Reasoning-Aware Table Pretraining
This addresses the problem of improving numerical reasoning in AI for table-based data analysis, representing a novel method for leveraging existing spreadsheet data.
The paper tackles the challenge of numerical reasoning over tables by using spreadsheet formulas as supervision for pretraining, achieving state-of-the-art results on cell type classification and formula prediction tasks.
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.