Correcting Semantic Parses with Natural Language through Dynamic Schema Encoding
This addresses the challenge of error correction in semantic parsing for database query systems, offering an incremental improvement over existing methods.
The paper tackles the problem of improving text-to-SQL semantic parsing by using natural language feedback to correct errors, boosting accuracy by up to 26% with one correction turn and enabling a smaller model to correct a larger one in a zero-shot setting.
In addressing the task of converting natural language to SQL queries, there are several semantic and syntactic challenges. It becomes increasingly important to understand and remedy the points of failure as the performance of semantic parsing systems improve. We explore semantic parse correction with natural language feedback, proposing a new solution built on the success of autoregressive decoders in text-to-SQL tasks. By separating the semantic and syntactic difficulties of the task, we show that the accuracy of text-to-SQL parsers can be boosted by up to 26% with only one turn of correction with natural language. Additionally, we show that a T5-base model is capable of correcting the errors of a T5-large model in a zero-shot, cross-parser setting.