Skip to content

[DOC-12608] Add Cosine Similarity Metric #290

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Nov 21, 2024
Merged

[DOC-12608] Add Cosine Similarity Metric #290

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8368cde
[DOC-12608] Adding another entry to similarity metric setting for cos…
sarahlwelton Nov 8, 2024
cb68479
[DOC-12608] Implementing feedback from SME review
sarahlwelton Nov 14, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 9 additions & 0 deletions modules/search/partials/vector-search-field-descriptions.adoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,15 @@ Smaller euclidean distances mean that the values of each coordinate in the vecto
+
It's best to use *l2_norm* similarity when your embeddings contain information about the count or measure of specific things, and your embedding model uses the same similarity metric.

* *cosine*: Calculated by adding the result of multiplying a vector's components, or the product of the magnitudes of the vectors and the cosine of the angle between them.
This metric is not affected by the size of the vectors being measured.
+
Use *cosine* similarity to get the best results with an embedding model that uses cosine similarity.
Cosine similarity works well for semantic search, document classification, and recommendation systems.
+
The Search Service will normalize any vectors before indexing when using cosine similarity.
Copy link

@abhinavdangeti abhinavdangeti Nov 13, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Worth mentioning that - we will also normalize the query vector prior to executing the nearest neighbor search if the field it is associated with is using the cosine similarity distance metric.

Use *dot_product* similarity if your vectors are already normalized.

For more information about Vector Search indexes, see xref:vector-search:vector-search.adoc[] or xref:vector-search:create-vector-search-index-ui.adoc[].
// end::similarity_metric[]
// tag::dimension[]
Expand Down