Build a database analysis tool that detects when the meaning of a field changes over time even though its name and technical type remain unchanged.
Accept schema definitions, sample records, migration files, data dictionaries, and versioned API payloads. Analyze changes in value distributions, null rates, units, enum values, naming conventions, and relationships between fields.
Identify possible semantic drift such as a price field changing from gross to net, a status field acquiring a new interpretation, timestamps changing timezone conventions, or a boolean flag being repurposed.
For each finding, show the affected versions, supporting evidence, downstream consumers, potential impact, and tests that could confirm the issue.
Generate data contracts, migration review questions, compatibility checks, and regression test suggestions. Provide a timeline that connects schema changes to observed data changes.
Do not infer semantic changes from statistical differences alone. Mark findings as hypotheses until verified, protect sensitive sample data, and make every recommendation traceable to the records or schema changes that motivated it.
0 Comments