You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An simple solution for BM25 indexing is to get all intersected documents through inverted index and than calculate score in a brute force way.
CREATEINDEXon document USING GIN(to_voc_ids(sparsevec) gin__int_ops)
# Then utilize GIN index to querySELECT pid FROM documents WHERE to_voc_ids(documents.embedding) && to_voc_ids(queries.embedding)
ORDER BYqueries.embedding<#> documents.embedding LIMIT 1
Probably should skip stop word index in to_voc_ids
The text was updated successfully, but these errors were encountered:
An simple solution for BM25 indexing is to get all intersected documents through inverted index and than calculate score in a brute force way.
Probably should skip stop word index in to_voc_ids
The text was updated successfully, but these errors were encountered: