auto_awesome Guided data journey

Turn a local file into a review-ready dataset.

Upload your table, let the hub inspect it, repair obvious structure problems, check it against the Hub quality standards and save a private draft for review.

Step 1

Drop a dataset file

The browser previews and prepares your data before anything is stored. CSV, TSV, JSON, JSONL and Excel files are supported.

upload_file
cloud_upload Drop your file here We will read it locally first, then you decide what to submit.
Checking provider access...

Step 2

Give the dataset a public-quality description

Keep this short. The reviewer can improve it later, but the dataset needs enough context to be understandable.

edit_note

Uploading never makes anything public: this only decides who can see the draft while you prepare it. Publishing in the catalogue always goes through the curation review.

You do not need to know the law: each card explains in plain words what you allow others to do with your data. Pick the one that sounds like you.

Who is providing these data?

Choose how the dataset should be attributed. You can submit personally, use one of your providers or request a new provider without leaving this upload.

Personal submission selected

The draft belongs to your account. You can link it to a provider later, before publication.

How provider attribution works

AttributionThe selected provider is credited on the public dataset.

Team accessIts authorised members can follow and manage the submission.

Safe by defaultNothing is published automatically; the first version remains private.

Type a country or region and pick it from the list — it is saved with its standard name (e.g. “españa” becomes Spain).
One year (2023) or a range (2020-2025), then Enter. Anything else is not accepted.

Step 3

Let the vocabulary understand your columns

The hub compares column names with the controlled vocabulary and highlights strong, weak and unresolved matches.

dataset
Column analysis has not run yet.

Step 4

Repair structure automatically

The repair engine normalizes headers and values where it can. You see the result before saving anything to the hub.

build_circle
Repair has not run yet.

Step 5

Run the quality standards and validation checks

This is the quality gate. The system will explain what looks ready and what still needs attention.

fact_check

Step 6

Save the private draft in the hub

The final click stores the dataset, version, tables, records, source file and generated reports in PostgreSQL. It is still private.

database_upload