CrateDB: Vector Store -- make _euclidean_relevance_score_fn identity f.

We don't need anything on top of it, ie we don't need this function and
instead should use value from CrateDB as is.

Similarity is already in the (0,1] interval and dividing by math.sqrt(2)
won't normalize it but return wrong result, for example 1 will become
0.714.

Author: amotl

libs/community/langchain_community/vectorstores/cratedb/base.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import enum
-import math
 from typing import (
     Any,
     Callable,
@@ -466,10 +465,10 @@ def _euclidean_relevance_score_fn(similarity: float) -> float:
         #  others are not!)
         # - embedding dimensionality
         # - etc.
-        # This function converts the euclidean norm of normalized embeddings
+        # This function converts the Euclidean norm of normalized embeddings
         # (0 is most similar, sqrt(2) most dissimilar)
         # to a similarity function (0 to 1)
 
         # Original:
         # return 1.0 - distance / math.sqrt(2)
-        return similarity / math.sqrt(2)
+        return similarity

libs/community/tests/integration_tests/vectorstores/test_cratedb.py

@@ -470,9 +470,9 @@ def test_cratedb_relevance_score() -> None:
     output = docsearch.similarity_search_with_relevance_scores("foo", k=3)
     # Original score values: 1.0, 0.9996744261675065, 0.9986996093328621
     assert output == [
-        (Document(page_content="foo", metadata={"page": "0"}), 0.7071067811865475),
-        (Document(page_content="bar", metadata={"page": "1"}), 0.35355339059327373),
-        (Document(page_content="baz", metadata={"page": "2"}), 0.1414213562373095),
+        (Document(page_content="foo", metadata={"page": "0"}), 1.0),
+        (Document(page_content="bar", metadata={"page": "1"}), 0.5),
+        (Document(page_content="baz", metadata={"page": "2"}), 0.2),
     ]