I am currently using atlas search to index all fields in my document except “keyword”. “Keyword” is indexed using atlas vector search.
The goal is to find documents by similarity to “keyword”, but at the same time filter by other fields.
For example, I want to find instagram account documents similar to hermes, but from France.
What would be the right way to construct my aggregation pipeline?
Hi @Isaac_Jieu! Thank you for the question.
We support prefiltering in vector search using the filter
field, which I believe would provide what you are looking for. You can filter on a variety of data types, which is expanding over time, and we are also exploring adding support for analyzed text and geospatial fields for more advanced prefiltering than what you can get with the MQL operators we currently support.