OptiSQL: Executable SQL Generation from Optical Tokens

TL;DR

OptiSQL is a vision-based framework that efficiently generates executable SQL directly from table images and natural language questions by using compact optical tokens, reducing token overhead and maintaining accuracy.

Contribution

This work introduces OptiSQL, a novel visual approach that compresses table information into optical tokens for SQL generation, addressing limitations of text-based schemas.

Findings

Retains strong execution accuracy with fewer tokens

Reduces table input tokens by an order of magnitude

Maintains structural information under visual perturbations

Abstract

Executable SQL generation is typically studied in text-to-SQL settings, where tables are provided as fully linearized textual schemas and contents. While effective, this formulation assumes access to structured text and incurs substantial token overhead, which is misaligned with many real-world scenarios where tables appear as visual artifacts in documents or webpages. We investigate whether compact optical representations can serve as an efficient interface for executable semantic parsing. We present OptiSQL, a vision-driven framework that generates executable SQL directly from table images and natural language questions using compact optical tokens. OptiSQL leverages an OCR-oriented visual encoder to compress table structure and content into a small set of optical tokens and fine-tunes a pretrained decoder for SQL generation while freezing the encoder to isolate representation sufficiency. Experiments on a visualized version of Spider 2.0-Snow show that OptiSQL retains strong execution accuracy while reducing table input tokens by an order of magnitude. Robustness analyses further demonstrate that optical tokens preserve essential structural information under visual perturbations.