Tip: Turn prodigy.db into web interface & JSON API with datasette

Just came across this new library on Twitter and immediately thought of all the cool possibilities for Prodigy. Haven't had time yet to play around with this extensively, but it looks promising, so I thought I'd post it here in case anyone wants to give it a go.

If you're using Prodigy with SQLite (the default database option), you should be able to simply point datasette at your prodigy.db, located in ~/.prodigy/prodigy.db by default. This lets you view your datasets and collected annotations in your browser, and even share them with others. It might also be very helpful for debugging custom recipes, and to find out what exactly is stored in the database.

:warning: NOTE: In order to only store each example once, the Prodigy SQLite database works with links that map example IDs to dataset IDs. This means that you can't get a nice, self-contained view of an individual dataset at this point. However, you can type in SQL queries!

3 Likes