BM25RetrievalStrategy

class langchain_elasticsearch.vectorstores.BM25RetrievalStrategy(k1: float | None = None, b: float | None = None)

Deprecated since version 0.2.0: Use BM25Strategy instead.

Retrieval strategy using the native BM25 algorithm of Elasticsearch.

Methods

__init__([k1, b])
before_index_setup(client, text_field, ...)	Executes before the index is created.
index(dims_length, vector_query_field, ...)	Executes when the index is created.
query(query_vector, query, k, fetch_k, ...)	Executes when a search is performed on the store.
require_inference()	Returns whether or not the strategy requires inference to be performed on the text before it is added to the index.

Parameters:

• k1 (float | None) –

• b (float | None) –

__init__(k1: float | None = None, b: float | None = None)

Parameters:

• k1 (float | None) –

• b (float | None) –

before_index_setup(client: Elasticsearch, text_field: str, vector_query_field: str) → None

Executes before the index is created. Used for setting up any required Elasticsearch resources like a pipeline.

Parameters:

• client (Elasticsearch) – The Elasticsearch client.

• text_field (str) – The field containing the text data in the index.

• vector_query_field (str) – The field containing the vector representations in the index.

Return type:

None

index(dims_length: int | None, vector_query_field: str, text_field: str, similarity: DistanceStrategy | None) → Dict

Executes when the index is created.

Parameters:

• dims_length (int | None) – Numeric length of the embedding vectors, or None if not using vector-based query.

• vector_query_field (str) – The field containing the vector representations in the index.

• text_field (str) – The field containing the text data in the index.

• similarity (DistanceStrategy | None) – The similarity strategy to use, or None if not using one.

Returns:

The Elasticsearch settings and mappings for the strategy.

Return type:

Dict

query(query_vector: List[float] | None, query: str | None, k: int, fetch_k: int, vector_query_field: str, text_field: str, filter: List[dict], similarity: DistanceStrategy | None) → Dict

Executes when a search is performed on the store.

Parameters:

• query_vector (List[float] | None) – The query vector, or None if not using vector-based query.

• query (str | None) – The text query, or None if not using text-based query.

• k (int) – The total number of results to retrieve.

• fetch_k (int) – The number of results to fetch initially.

• vector_query_field (str) – The field containing the vector representations in the index.

• text_field (str) – The field containing the text data in the index.

• filter (List[dict]) – List of filter clauses to apply to the query.

• similarity (DistanceStrategy | None) – The similarity strategy to use, or None if not using one.

Returns:

The Elasticsearch query body.

Return type:

Dict

require_inference() → bool

Returns whether or not the strategy requires inference to be performed on the text before it is added to the index.

Returns:

Whether or not the strategy requires inference to be performed on the text before it is added to the index.

Return type:

bool