Our research focuses on Data+AI, bridging the gap between data management and artificial intelligence. We explore two main directions: DB4AI (Database for AI) and AI4DB (AI for Database).
Below is a list of our specific projects and publications in these areas.
We design next-generation data systems tailored for AI applications, supporting multimodal data and complex analytical queries.
We build intelligent data agents and systems capable of handling semantic and analytical queries over heterogeneous data sources.
We develop efficient indexing and search algorithms to support hybrid queries over vector and relational data.
We optimize LSM-tree based storage engines to efficiently manage multimodal data on modern hardware.
We apply machine learning technqiues to enhance the performance and manageability of database systems.
We explore learning-based query optimization, focusing on cardinality estimation, plan representation, and rewrite systems to surpass traditional optimizers.
We investigate the design of learned index structures that adapt to data distribution and workload patterns for better efficiency.