GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing

Tao Yu · Chien-Sheng Wu · Xi V Lin · bailin wang · Yi Tan · Xinyi Yang · Dragomir Radev · Richard Socher · Caiming Xiong

Keywords: [ semantic parsing ] [ text-to-sql ] [ pre-training ] [ nlp ]

[ Abstract ]
[ Paper ]
Mon 3 May 9 a.m. PDT — 11 a.m. PDT


We present GraPPa, an effective pre-training approach for table semantic parsing that learns a compositional inductive bias in the joint representations of textual and tabular data. We construct synthetic question-SQL pairs over high-quality tables via a synchronous context-free grammar (SCFG). We pre-train our model on the synthetic data to inject important structural properties commonly found in semantic parsing into the pre-training language model. To maintain the model's ability to represent real-world data, we also include masked language modeling (MLM) on several existing table-related datasets to regularize our pre-training process. Our proposed pre-training strategy is much data-efficient. When incorporated with strong base semantic parsers, GraPPa achieves new state-of-the-art results on four popular fully supervised and weakly supervised table semantic parsing tasks.

Chat is not available.