Query Documents in a Collection

Query Documents in a Collection

Usage

query(
  client,
  collection_name,
  query_embeddings,
  n_results = 10L,
  where = NULL,
  where_document = NULL,
  include = c("documents", "metadatas", "distances"),
  tenant = "default_tenant",
  database = "default_database"
)

Arguments

client

A ChromaDB client object

collection_name

Name of the collection

query_embeddings

List of query embeddings (must be a list of numeric vectors)

n_results

Number of results to return per query (default: 10)

where

Optional filtering conditions

where_document

Optional document-based filtering conditions

include

Optional vector of what to include in results. Possible values: "documents", "embeddings", "metadatas", "distances", "uris", "data" (default: c("documents", "metadatas", "distances"))

tenant

The tenant name (default: "default")

database

The database name (default: "default")

Value

A list containing the query results. Each element (documents, metadatas, distances) is a nested list, so use double brackets [[]] to access individual elements.

Details

Note that ChromaDB's API only accepts embeddings for queries. If you want to query using text, you need to first convert your text to embeddings using an embedding model (e.g., using OpenAI's API, HuggingFace's API, or a local model).

Example:

# First convert text to embeddings using your preferred method
text_embedding <- your_embedding_function("your search text")
# Then query using the embedding
result <- query(client, "my_collection",
               query_embeddings = list(text_embedding))