A searchable data catalogue from scattered descriptions

The situation

Dataset descriptions arrived by email and in other forms from eight data stewards in different supplier organisations. Reviewing and updating the catalogue by hand took so long that it often did not happen, so people could not find the data that existed.

What we did

We automated the update. Every change is now suggested automatically, from the incoming descriptions, for a person to review and accept. The result is a discoverable catalogue of metadata.

What changed

Review and update now take 80% less time than doing it by hand, so the catalogue keeps up.

Why it matters

People stay in charge of what goes into the catalogue, but no longer do the typing. A catalogue that is current is one people actually use.