Vector collection with indexing and search
nameCollection name
dimensionVector dimension
metricDistance metric
languageLanguage setting ("en" or "ml")
new()Create a new Collection
Collection$new(
name,
dimension,
metric = "cosine",
storage = NULL,
language = "en"
)nameCollection name
dimensionVector dimension
metricDistance metric
storageStorage backend
languageLanguage behavior ("en" = ASCII-focused, "ml" = Unicode-aware)
add()Add documents to collection
Collection$add(ids, vectors, metadata = NULL, texts = NULL)idsDocument IDs
vectorsMatrix of vectors
metadataList of metadata
textsCharacter vector of texts
search()Search collection
Collection$search(query, limit = 10, filter = NULL, include_vectors = FALSE)queryQuery vector
limitNumber of results
filterMetadata filter
include_vectorsInclude vectors in results
Results object
keyword_search()Keyword search
Collection$keyword_search(query_text, limit = 10, filter = NULL)query_textQuery text
limitNumber of results
filterMetadata filter
Results object
hybrid_search()Hybrid search (dense + sparse)
Collection$hybrid_search(
query,
query_text,
limit = 10,
vector_weight = 0.5,
text_weight = 0.5,
filter = NULL,
include_vectors = FALSE,
rrf_k = 60,
prefetch_multiplier = 10
)queryQuery vector
query_textQuery text
limitNumber of results
vector_weightWeight for vector search
text_weightWeight for text search
filterMetadata filter
include_vectorsInclude vectors in results
rrf_kRRF constant
prefetch_multiplierPrefetch multiplier
Results object
idsDocument IDs
List of results
delete()Delete documents by ID
Collection$delete(ids)idsDocument IDs to delete
count()Get document count
Collection$count()Integer count
clear()Clear collection
Collection$clear()
clone()The objects of this class are cloneable with this method.
Collection$clone(deep = FALSE)deepWhether to make a deep clone.