Use cases¶

For data analysts and scientists¶

• Express cleaning, deduplication and linking as one plan you can iterate on, instead of a pile of notebook cells
• Measure whether a change actually improved the match, against judgements or known truth
• Re-run cheaply. Collecting again only redoes the work whose inputs changed

Install

For data architects and engineers¶

• Reconcile entities across disparate sources without moving the data anywhere
• Compare methodologies on equal terms rather than by argument
• Take the plan an analyst built and run it operationally. It's the same object
• Provide the entities a semantic layer is built from

Build a plan

For service owners¶

• Understand the broader business entities in your service, not just the records you hold
• Resolve a single key against every other source you match against, offline and fast
• Keep the output stable enough for other services to depend on

Query the result