-
Type: Bug
-
Resolution: Duplicate
-
Priority: Minor - P4
-
Affects Version/s: 1.40.3
-
Component/s: None
-
None
-
Environment:OS: MacOS Apple Silicon
node.js / npm versions: 20
Additional info: MongoDB Compass Aggregations / MongoDB A.I. Docs
-
Not Needed
Problem Statement/Rationale
What is going wrong? What action would you like the Engineering team to take?
I'm receiving conflicting answers from the A.I. document chat vs. the aggregation pipeline. I'm told by the index type needs to be knnVector by MongoDB AI, but in Compass's stage dropdown in Aggregations it states:
```
$vectorSearch - Atlas only. Performs a kNN search on a vector in the specified field or fields. The field that you want to search must be covered by an Atlas Search index of type `vectorEmbedding`.
```
However, MongoDB AI said:
"""
No, knnVector is not covered by vectorEmbedding in Atlas Search. When using $vectorSearch in a pipeline, the fields to search must be indexed as knnVector type. However, MongoDB Compass may suggest that the field to search must be covered by an Atlas Search index of type vectorEmbedding. This is incorrect.
Further reading:
- https://mongodb.com/docs/atlas/atlas-search/field-types/knn-vector/
- https://mongodb.com/docs/atlas/atlas-search/knn-beta/
- https://mongodb.com/docs/atlas/atlas-vector-search/vector-search-stage/
"""
Also, it looks like on https://mongodb.com/docs/atlas/atlas-vector-search/vector-search-stage/ it says:
"""
The $vectorSearch stage performs an aNN search on a vector in the specified field. The field that you want to search must be indexed as Atlas Vector Search knnVector type.
"""
Steps to Reproduce
How could an engineer replicate the issue you’re reporting?
Open the Aggregations tab for a collection in Compass, click on 'Stage 1', scroll down to '$vectorSearch'.
- duplicates
-
COMPASS-7300 Fix vector search index type name in template
- Closed
- related to
-
COMPASS-7288 Use knnVector instead of vectorEmbedding in index template definition
- Closed