Table Retrieval using LLMs and Semantic Table Similarity

Searching for relevant tables in response to a textual phrase or a question is an important task for large tabular data repositories, such as relational databases, CSV files in data lakes, etc. It is somewhat different from the problem of web document search because the subjects of search are tables rather than documents, while the query remains textual. In this paper, we explore a novel technique for table search on large repositories using natural language queries. It is based on a generative methodology that aims to maximize the semantic connection between the query and the resulting tables. Unlike traditional keyword search approaches, our technique can find the needed tables more effectively through deeper semantic concept discovery rather than simply searching for exact keyword matches. Additionally, our technique supports natural language queries rather than plain keyword queries. In this paper, we describe the core ideas, implementation, and effectiveness of our method using two different benchmarks with diverse queries.