Leveraging Data Recasting to Enhance Tabular Reasoning
This work addresses the need for scalable and linguistically diverse data in tabular reasoning, offering a practical solution for researchers in natural language processing and AI, though it is incremental as it builds on existing data generation strategies.
The authors tackled the problem of generating challenging tabular inference data by developing a semi-automatic framework that recasts existing datasets, combining the benefits of human annotation and synthetic generation. They used this framework to create tabular NLI instances from five datasets, showing it can serve as evaluation benchmarks and augmentation data to enhance performance on tabular NLI tasks, with analysis of zero-shot effectiveness and performance trends across dataset types.
Creating challenging tabular inference data is essential for learning complex reasoning. Prior work has mostly relied on two data generation strategies. The first is human annotation, which yields linguistically diverse data but is difficult to scale. The second category for creation is synthetic generation, which is scalable and cost effective but lacks inventiveness. In this research, we present a framework for semi-automatically recasting existing tabular data to make use of the benefits of both approaches. We utilize our framework to build tabular NLI instances from five datasets that were initially intended for tasks like table2text creation, tabular Q/A, and semantic parsing. We demonstrate that recasted data could be used as evaluation benchmarks as well as augmentation data to enhance performance on tabular NLI tasks. Furthermore, we investigate the effectiveness of models trained on recasted data in the zero-shot scenario, and analyse trends in performance across different recasted datasets types.