Abstract
Tabular question answering (TQA) presents a challenging setting for neural systems by requiring joint reasoning of natural language with large amounts of semi-structured data. Unlike humans who use programmatic tools like filters to transform data before processing, language models in TQA process tables directly, resulting in information loss as table size increases. In this paper we propose ToolWriter to generate query specific programs and detect when to apply them to transform tables and align them with the TQA model’s capabilities. Focusing Toolwriter to generate row-filtering tools improves the state-of-the-art for WikiTableQuestions and WikiSQL with the most performance gained on long tables. By investigating headroom, our work highlights the broader potential for programmatic tools combined with neural components to manipulate large amounts of structured data.- Anthology ID:
- 2023.emnlp-main.1003
- Volume:
- Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 16137–16148
- Language:
- URL:
- https://aclanthology.org/2023.emnlp-main.1003
- DOI:
- 10.18653/v1/2023.emnlp-main.1003
- Cite (ACL):
- Carlos Gemmell and Jeff Dalton. 2023. ToolWriter: Question Specific Tool Synthesis for Tabular Data. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 16137–16148, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- ToolWriter: Question Specific Tool Synthesis for Tabular Data (Gemmell & Dalton, EMNLP 2023)
- PDF:
- https://preview.aclanthology.org/corrections-2024-05/2023.emnlp-main.1003.pdf