Volta at SemEval-2021 Task 9: Statement Verification and Evidence Finding with Tables using TAPAS and Transfer Learning

Tables are widely used in various kinds of documents to present information\nconcisely. Understanding tables is a challenging problem that requires an\nunderstanding of language and table structure, along with numerical and logical\nreasoning. In this paper, we present our systems to solve Task 9 of\nSemEval-2021: Statement Verification and Evidence Finding with Tables\n(SEM-TAB-FACTS). The task consists of two subtasks: (A) Given a table and a\nstatement, predicting whether the table supports the statement and (B)\nPredicting which cells in the table provide evidence for/against the statement.\nWe fine-tune TAPAS (a model which extends BERT's architecture to capture\ntabular structure) for both the subtasks as it has shown state-of-the-art\nperformance in various table understanding tasks. In subtask A, we evaluate how\ntransfer learning and standardizing tables to have a single header row improves\nTAPAS' performance. In subtask B, we evaluate how different fine-tuning\nstrategies can improve TAPAS' performance. Our systems achieve an F1 score of\n67.34 in subtask A three-way classification, 72.89 in subtask A two-way\nclassification, and 62.95 in subtask B.\n