AI glossary · Basic · Data engineering
Data catalog
A data catalog is an organized list and search service that lets people look up, in one place, what data exists in a company, where it is, what it means, who manages it and whether it can be trusted.
About this term
- Level
- Basic · Technical terms you often meet in AI news
- Category
- Data, ops & platforms
- Fields
- Data engineering
- Added
- 2026-10-09
How it works
A data catalog collects not the data itself but information about the data, that is, metadata. It automatically scans databases, data lakes, warehouses and so on to gather technical information such as table names, column structure and update times. People then add business descriptions such as 'revenue in this table includes VAT', the owner, and tags such as whether it contains sensitive information. Users type something like 'monthly revenue' into the search box to find related data, check its description and quality information, and then use it.
How a catalog is built and used
- 1Automatic collectionTable names, column structure, update times
- 2Adding descriptionsBusiness meaning, owner, sensitive-data tags
- 3SearchLook up e.g. 'monthly revenue'
- 4Check, then useDecide from description and quality
Why it matters
As a company's data grows, a lot of time goes into asking 'Does this data even exist?' or 'There are several similar tables; which one is right?' With a catalog, analysts and AI developers can find data on their own without asking owners one by one. Flagging data that contains personal information helps prevent careless use, making the catalog a foundation of data governance. It is also a key tool for keeping a data lake from turning into a 'data swamp'.
Things to know
Keeping a catalog up to date is harder than building it. Technical information can be collected automatically, but information such as business meaning and owners must be filled in by people on an ongoing basis, so it is easily neglected. That is why catalogs are often connected to data lineage, which shows where data comes from and where it flows, and to quality check results, to make them a 'living' list. Recently LLMs have also been used to draft table and column descriptions automatically, and features such as Text-to-SQL refer to catalog descriptions to improve accuracy.
Example
Picture a data analyst at a financial company who wants to build a 'customer churn prediction model' and searches for 'cancellation' in the company's data catalog. In the results they find a table of cancellation history, check in the description the meaning of each column, the responsible department and the personal-information flag, then request access and start the analysis right away.