LIMA: Denial Constraint Discovery in Large Dynamic Databases

Modern Database Management Systems rely on metadata, such as data profiles and attribute associations, to function effectively. However, generating and maintaining this metadata on very large databases often incurs significant computational and storage costs. In this demonstration paper, we present a system capable of efficiently inferring and maintaining metadata in the form of Denial Constraints (DCs). This formalism has attracted considerable interest due to its expressiveness, as it can capture a wide range of data rules, including Functional Dependencies, Keys, and Order Dependencies. Rather than repeatedly extracting DCs whenever they are needed, our system maintains this metadata continuously and updates it seamlessly as the underlying data evolves, with minimal computational and memory overhead. This is the first system that enables the practical discovery of DCs on very large datasets, thereby opening the door to numerous applications that were previously infeasible due to the high cost of obtaining DCs.