Add duplicate filtering by document ID in HNSWlib search #623
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit modifies HNSWlib to filter duplicate document IDs during KNN search, ensuring only one embedding per unique document ID is returned. Key changes include: - Added `internal_id_to_doc_id_` vector to `HierarchicalNSW` to map internal IDs to document IDs, populated in `addPoint`. - Introduced `getMetadata` method to retrieve document IDs. - Extended `VisitedList` with `seen_doc_ids` set to track seen document IDs thread-locally, avoiding mutex contention. - Updated `searchBaseLayerST` to skip candidates with already-seen document IDs using `vl->is_doc_seen(doc_id)`. - Removed unused `visited_metadata_` and `visited_metadata_lock_` as filtering is now handled by `VisitedList`. The duplicate filtering works as intended, though `knnQuery` may raise a `RuntimeError` if `k` exceeds the number of unique document IDs due to result array shape constraints. Tests for basic filtering, single ID, and large datasets pass, while empty index and insufficient IDs cases require further handling. Files modified: - hnswalg.h: Added duplicate filtering logic and mappings. - visited_list_pool.h: Enhanced `VisitedList` for document ID tracking.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit modifies HNSWlib to filter duplicate document IDs during KNN search, ensuring only one embedding per unique document ID is returned. Key changes include:
internal_id_to_doc_id_vector toHierarchicalNSWto map internal IDs to document IDs, populated inaddPoint.getMetadatamethod to retrieve document IDs.VisitedListwithseen_doc_idsset to track seen document IDs thread-locally, avoiding mutex contention.searchBaseLayerSTto skip candidates with already-seen document IDs usingvl->is_doc_seen(doc_id).visited_metadata_andvisited_metadata_lock_as filtering is now handled byVisitedList. The duplicate filtering works as intended, thoughknnQuerymay raise aRuntimeErrorifkexceeds the number of unique document IDs due to result array shape constraints. Tests for basic filtering, single ID, and large datasets pass, while empty index and insufficient IDs cases require further handling.Files modified:
VisitedListfor document ID tracking.