Skip to content

Fix bug where intercepted semantic knn queries did not respect filters #121410

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 1 commit into from
Jan 31, 2025
Merged

Fix bug where intercepted semantic knn queries did not respect filters #121410

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
4 changes: 3 additions & 1 deletion ...k/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/InferenceFeatures.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@

import java.util.Set;

import static org.elasticsearch.xpack.inference.queries.SemanticKnnVectorQueryRewriteInterceptor.SEMANTIC_KNN_FILTER_FIX;
import static org.elasticsearch.xpack.inference.queries.SemanticKnnVectorQueryRewriteInterceptor.SEMANTIC_KNN_VECTOR_QUERY_REWRITE_INTERCEPTION_SUPPORTED;
import static org.elasticsearch.xpack.inference.queries.SemanticMatchQueryRewriteInterceptor.SEMANTIC_MATCH_QUERY_REWRITE_INTERCEPTION_SUPPORTED;
import static org.elasticsearch.xpack.inference.queries.SemanticSparseVectorQueryRewriteInterceptor.SEMANTIC_SPARSE_VECTOR_QUERY_REWRITE_INTERCEPTION_SUPPORTED;
Expand Down Expand Up @@ -42,7 +43,8 @@ public Set<NodeFeature> getTestFeatures() {
SEMANTIC_KNN_VECTOR_QUERY_REWRITE_INTERCEPTION_SUPPORTED,
TextSimilarityRankRetrieverBuilder.TEXT_SIMILARITY_RERANKER_ALIAS_HANDLING_FIX,
SemanticInferenceMetadataFieldsMapper.INFERENCE_METADATA_FIELDS_ENABLED_BY_DEFAULT,
SEMANTIC_TEXT_HIGHLIGHTER_DEFAULT
SEMANTIC_TEXT_HIGHLIGHTER_DEFAULT,
SEMANTIC_KNN_FILTER_FIX
);
}
}
10 changes: 8 additions & 2 deletions ...a/org/elasticsearch/xpack/inference/queries/SemanticKnnVectorQueryRewriteInterceptor.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,7 @@ public class SemanticKnnVectorQueryRewriteInterceptor extends SemanticQueryRewri
public static final NodeFeature SEMANTIC_KNN_VECTOR_QUERY_REWRITE_INTERCEPTION_SUPPORTED = new NodeFeature(
"search.semantic_knn_vector_query_rewrite_interception_supported"
);
public static final NodeFeature SEMANTIC_KNN_FILTER_FIX = new NodeFeature("search.semantic_knn_filter_fix");
Copy link
Contributor

Choose a reason for hiding this comment

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

is this value semantic_knn_filter_fix defined somewhere, or does the NodeFeature require a unique string to identify the fix from a specific version?

Copy link
Member Author

Choose a reason for hiding this comment

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

It's required for testing, so we only test the fix on versions that have the feature. This is the pattern we use for defining node features.


public SemanticKnnVectorQueryRewriteInterceptor() {}

Expand Down Expand Up @@ -147,6 +148,7 @@ private KnnVectorQueryBuilder addIndexFilterToKnnVectorQuery(Collection<String>
);
}

copy.addFilterQueries(original.filterQueries());
Copy link
Contributor

Choose a reason for hiding this comment

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

Curious—how do we ensure that the same filter is not added twice? Also, do we handle a different type of filter using copy.addFilterQuery(new TermsQueryBuilder(IndexFieldMapper.NAME, indices))?

Copy link
Member Author

Choose a reason for hiding this comment

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

This is adding the filters that were already added to the request, we do not need to do further analysis as we're translating the query. Adding the filter on index name allows us to filter to only those indices we want to query.

Copy link
Contributor

Choose a reason for hiding this comment

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

Adding the filter on index name allows us to filter to only those indices we want to query I missed this key point. Thank you for the clarifications.

copy.addFilterQuery(new TermsQueryBuilder(IndexFieldMapper.NAME, indices));
return copy;
}
Expand All @@ -165,16 +167,17 @@ private KnnVectorQueryBuilder buildNewKnnVectorQuery(
KnnVectorQueryBuilder original,
QueryVectorBuilder queryVectorBuilder
) {
KnnVectorQueryBuilder newQueryBuilder;
if (original.queryVectorBuilder() != null) {
return new KnnVectorQueryBuilder(
newQueryBuilder = new KnnVectorQueryBuilder(
fieldName,
queryVectorBuilder,
original.k(),
original.numCands(),
original.getVectorSimilarity()
);
} else {
return new KnnVectorQueryBuilder(
newQueryBuilder = new KnnVectorQueryBuilder(
fieldName,
original.queryVector(),
original.k(),
Expand All @@ -183,6 +186,9 @@ private KnnVectorQueryBuilder buildNewKnnVectorQuery(
original.getVectorSimilarity()
);
}

newQueryBuilder.addFilterQueries(original.filterQueries());
return newQueryBuilder;
}

@Override
Expand Down
40 changes: 40 additions & 0 deletions ...nference/src/yamlRestTest/resources/rest-api-spec/test/inference/47_semantic_text_knn.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,6 +43,8 @@ setup:
body:
mappings:
properties:
keyword_field:
type: keyword
inference_field:
type: semantic_text
inference_id: dense-inference-id
Expand All @@ -53,6 +55,8 @@ setup:
body:
mappings:
properties:
keyword_field:
type: keyword
inference_field:
type: semantic_text
inference_id: dense-inference-id-2
Expand All @@ -63,6 +67,8 @@ setup:
body:
mappings:
properties:
keyword_field:
type: keyword
inference_field:
type: dense_vector
dims: 10
Expand All @@ -74,6 +80,8 @@ setup:
body:
mappings:
properties:
keyword_field:
type: keyword
inference_field:
type: dense_vector
dims: 3
Expand All @@ -84,6 +92,7 @@ setup:
index: test-semantic-text-index
id: doc_1
body:
keyword_field: "foo"
inference_field: [ "inference test", "another inference test" ]
refresh: true

Expand All @@ -92,6 +101,7 @@ setup:
index: test-semantic-text-index-2
id: doc_2
body:
keyword_field: "bar"
inference_field: [ "inference test", "another inference test" ]
refresh: true

Expand All @@ -100,6 +110,7 @@ setup:
index: test-dense-vector-index
id: doc_3
body:
keyword_field: "baz"
inference_field: [ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 ]
refresh: true

Expand All @@ -108,6 +119,7 @@ setup:
index: test-incompatible-dense-vector-index
id: doc_4
body:
keyword_field: "qux"
inference_field: [ 1, 2, 3 ]
refresh: true

Expand Down Expand Up @@ -311,6 +323,34 @@ setup:

- match: { hits.total.value: 2 }

---
"knn query respects filters":
- requires:
cluster_features: "search.semantic_knn_filter_fix"
reason: filters fixed in 8.18.0

- do:
search:
index:
- test-semantic-text-index
- test-semantic-text-index-2
body:
query:
knn:
field: inference_field
k: 10
num_candidates: 100
query_vector_builder:
text_embedding:
model_text: test
filter:
term:
keyword_field: "foo"

- match: { hits.total.value: 1 }
- match: { hits.hits.0._id: "doc_1" }



---
"knn query against multiple semantic_text fields with multiple inference IDs specified in semantic_text fields with smaller k returns k for each index":
Expand Down