Constraint-Driven Database Repair

Synonyms

Data reconciliation; Data standardization; Minimal-change integrity maintenance

Definition

Given a set Σ of integrity constraints and a database instance D of a schema R, the problem of constraint-driven database repair is to find an instance D^′ of the same schema R such that (i) D^′ is consistent, i.e., D^′ satisfies Σ, and moreover, (ii) D^′minimally differs from the original database D, i.e., it takes a minimal number of repair operations or incurs minimal cost to obtain D^′ by updating D.

Historical Background

Real-life data is often dirty, i.e., inconsistent, inaccurate, stale, or deliberately falsified. While the prevalent use of the Web has made it possible, on an unprecedented scale, to extract and integrate data from diverse sources, it has also increased the risks of creating and propagating dirty data. Dirty data routinely leads to misleading or biased analytical results and decisions and incurs loss of revenue, credibility, and customers. With this comes the need for...