Currently, I'm using the default SQLite database to store prodigy dataset. As my documents are stored in Neo4j, I wonder can I link prodigy to Neo4j database? If not, is there any other options? Thanks!
Yes, that should be no problem!
If you want to store your data differently, you can always implement a custom
Database class that implements the same method's as Prodigy's
Database wrapper – for instance, a
get_dataset method that takes a dataset name and returns the annotations in the dataset, an
add_examples method that adds examples to one or more datasets, and so on. Within those methods, you can then handle the saving and loading however you want. Also see here for details: https://prodi.gy/docs/api-database#custom
You can then provide that custom database the
"db" returned by a custom recipe or expose it via an entry point (which would also let you write something like
"db": "neo4j" in your Prodigy config).