SIGMOD Research Awards for Project on Cardinality Estimation
A recent breakthrough on the longstanding problem of cardinality estimation in database systems brought the DaST team and collaborators two major awards in data management research.
Cardinality estimation is dubbed the Achilles' heel of query optimizers: Given a database query, the goal is to estimate the size of its output without computing the query. Query optimizers rely on these estimates to decide how to execute a query. Inaccurate estimates can lead to poor query plans and significant runtime performance penalties.
The DaST team and collaborators introduced a new approach to cardinality estimation that significantly furthers the state-of-the-art theory and systems on this problem. Their approach is rooted in information theory and computes a guaranteed upper bound on the cardinality of the query output. In practice, it is shown to be close to the true cardinality.
Their theoretical work on cardinality estimation, which appeared in the Proceedings of the ACM on Management of Data (PODS) 2024, received the SIGMOD Research Highlights Award.
Their systems work on cardinality estimation, which will appear in the Proceedings of the ACM on Management of Data (SIGMOD) 2025, received the Best Paper Award.
The ACM conferences PODS and SIGMOD are the internationally top conferences on the theory and respectively systems of data management.
Project page with papers: https://www.ifi.uzh.ch/en/dast/research/LpBound.html