Update CHANGELOG

Author: willtai

CHANGELOG.md

@@ -11,7 +11,7 @@
 - Introduced Neo4jMessageHistory and InMemoryMessageHistory classes for managing LLM message histories.
 - Added examples and documentation for using message history with Neo4j and in-memory storage.
 - Updated LLM and GraphRAG classes to support new message history classes.
-
+- Introduced a linear hybrid search ranker for HybridRetriever and HybridCypherRetriever, allowing customizable ranking with an `alpha` parameter.
 ### Changed
 
 - Refactored index-related functions for improved compatibility and functionality.

examples/retrieve/hybrid_retriever.py

@@ -37,7 +37,7 @@
     # Perform the similarity search for a text query
     # (retrieve the top 5 most similar nodes)
     query_text = "Find me a movie about aliens"
-    print(retriever.search(query_text=query_text, top_k=5))
+    print(retriever.search(query_text=query_text, top_k=5)
 
     # note: it is also possible to query from a query_vector directly:
     # query_vector: list[float] = [...]

src/neo4j_graphrag/retrievers/hybrid.py

@@ -198,7 +198,6 @@ def get_search_results(
                 )
             query_vector = self.embedder.embed_query(query_text)
             parameters["query_vector"] = query_vector
-
         search_query, _ = get_search_query(
             search_type=SearchType.HYBRID,
             return_properties=self.return_properties,

src/neo4j_graphrag/types.py

@@ -26,6 +26,7 @@
     field_validator,
     model_validator,
 )
+from typing_extensions import Self
 
 from neo4j_graphrag.utils.validation import validate_search_query_input
 
@@ -169,12 +170,12 @@ def validate_ranker(cls, v: Union[str, HybridSearchRanker]) -> HybridSearchRanke
             allowed = ", ".join([r.value for r in HybridSearchRanker])
             raise ValueError(f"Invalid ranker type. Allowed values are: {allowed}.")
 
-    @model_validator(mode="before")
-    def validate_alpha(cls, values: dict[str, Any]) -> dict[str, Any]:
-        ranker, alpha = values.get("ranker"), values.get("alpha")
+    @model_validator(mode="after")
+    def validate_alpha(self) -> Self:
+        ranker, alpha = self.ranker, self.alpha
         if ranker == HybridSearchRanker.LINEAR:
             if alpha is None:
-                values["alpha"] = 0.5
+                self.alpha = 0.5
             if isinstance(alpha, float) and not (0.0 <= alpha <= 1.0):
                 raise ValueError("alpha must be between 0 and 1")
         else:
@@ -183,8 +184,8 @@ def validate_alpha(cls, values: dict[str, Any]) -> dict[str, Any]:
                     "alpha parameter is only used when ranker is 'linear'. Ignoring alpha.",
                     UserWarning,
                 )
-                values["alpha"] = None
-        return values
+                self.alpha = None
+        return self
 
 
 class HybridCypherSearchModel(HybridSearchModel):