S$^2$SQL: Injecting Syntax to Question-Schema Interaction Graph Encoder for Text-to-SQL Parsers

The task of converting a natural language question into an executable SQL\nquery, known as text-to-SQL, is an important branch of semantic parsing. The\nstate-of-the-art graph-based encoder has been successfully used in this task\nbut does not model the question syntax well. In this paper, we propose\nS$^2$SQL, injecting Syntax to question-Schema graph encoder for Text-to-SQL\nparsers, which effectively leverages the syntactic dependency information of\nquestions in text-to-SQL to improve the performance. We also employ the\ndecoupling constraint to induce diverse relational edge embedding, which\nfurther improves the network's performance. Experiments on the Spider and\nrobustness setting Spider-Syn demonstrate that the proposed approach\noutperforms all existing methods when pre-training models are used, resulting\nin a performance ranks first on the Spider leaderboard.\n