User-Specific Data Separation (multi tenancy) and/or filtering.

[pXius]

I'm working on a project that requires user-specific data to be properly separated or, at the very least, filtered when retrieved. This includes ensuring that neighboring data does not affect rankings, links, or other relationships.

Given that databases like Postgres and Mongo support this in different ways, is there a way to leverage these capabilities in Morphik today? Or would this require additional implementations or integrations?

Would appreciate any insights on how best to approach this.

[ArnavAgrawal03]

TDLR: We support this! You can achieve this by tagging user ids at ingestion time, and filtering based on that at retrieval time.

Thank you for your question!

We support filtering before retrieval. That is, as long as you've added a user_id or something of the like in your metadata when you ingest user data, you can filter based on that at retrieval or query time.

The filtering takes place before search, and so filtering will not affect ranking, links, or other relationships. In fact, it may make search faster in cases where computing similarity is a speed bottleneck.

[pXius]

Amazing!
Going to give it a shot.
Thanks 😎

[Adityav369]

@pXius we added folder and user scoping now, so it's easier to manage users and keep data separate: https://docs.morphik.ai/concepts/user-folder-scoping