Error Detection for Text-to-SQL Semantic Parsing
This addresses trustworthiness issues for users deploying text-to-SQL systems, though it is incremental as it builds on existing parsing methods.
The paper tackles the problem of overconfidence in text-to-SQL parsers by proposing a parser-independent error detection model that uses a language model of code and graph neural networks, achieving improved performance over parser-dependent uncertainty metrics in experiments with three parsers.
Despite remarkable progress in text-to-SQL semantic parsing in recent years, the performance of existing parsers is still far from perfect. Specifically, modern text-to-SQL parsers based on deep learning are often over-confident, thus casting doubt on their trustworthiness when deployed for real use. In this paper, we propose a parser-independent error detection model for text-to-SQL semantic parsing. Using a language model of code as its bedrock, we enhance our error detection model with graph neural networks that learn structural features of both natural language questions and SQL queries. We train our model on realistic parsing errors collected from a cross-domain setting, which leads to stronger generalization ability. Experiments with three strong text-to-SQL parsers featuring different decoding mechanisms show that our approach outperforms parser-dependent uncertainty metrics. Our model could also effectively improve the performance and usability of text-to-SQL semantic parsers regardless of their architectures. (Our implementation is available at https://github.com/OSU-NLP-Group/Text2SQL-Error-Detection)