Paper accepted to CIDR 2024

CAESURA: Language Models as Multi-Modal Query Planners


Authors: Matthias Urban, Carsten Binnig

In this paper, we propose Language-Model-Driven Query Planning which allows the translation of natural language queries to query plans that can process arbitrary modalities. Our GPT-4 based prototype shows the great potential of this idea on two multi-modal datasets, signaling a promising step towards truly intelligent data systems. A preprint of the paper can be found here: link