deduplicate_chunks

deduplicate_chunks(bucket_ids, doc_ids)

Boolean mask keeping one representative per (bucket, doc) pair.

Multi-topic documents retain multiple representatives (one per bucket they touch). Single-topic documents collapse to fewer points.

Parameters

Name	Type	Description	Default
bucket_ids		Array-like of bucket assignments per point.	required
doc_ids		Array-like of document identifiers per point (same length).	required

Returns

Name	Type	Description
	np.ndarray	Boolean array of length len(bucket_ids). True for representative points.