Scalable Data Integration with LLMs

This repository contains two seperate (but related) research projects on using LLMs to generate transformations between two schemas, reconciling their structual differences. Both projects share the same application, but focus on different levels of complexity in terms of the expressivness of their generated programs. As such, they also require different techniques and datasets. The Stable Schema Matching project focuses on generating column alignments across two schemas (i.e., pairs of semantically similar columns), while the Schema Mapping project focuses on generating queries. More details as well as datasets and code for each project can be found in their respective directories.