I want to check an algorithm that learns the connections between different tables in a database.

Where can I find datasets that are compound of multiple tables (i.e. more than 5) that have connections between them?

All I can find on the web (mainly on Kaggle.com) are datasets with 1 or 2 tables.

I'm aware that my question is kind of off topic but still I'm sure it can help people that have the same problem as I do


There are many places:

this one is very good: https://relational.fit.cvut.cz/

there are also:




Is there any reason why you couldn't create them yourself? I assume your algorithm is doing things like looking for primary and foreign keys, so the actual data in the tables shouldn't really matter.

I would treat this as an exercise in data augmentation and just synthesize a series of tables with all kinds of keys. Some would have integer keys, others GUIDs, others some sort of hash column, etc, etc. It would be fairly straightforward to create a script that can create you any number of tables with matching keys like this.

  • $\begingroup$ The algorithm ignores key constraints and looks only on the data, I can synthesize data for a development environment, but real-world data is much better for the end to end tests. $\endgroup$ – yossico Jan 10 '19 at 9:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.