feat: Add configurable embedding similarity threshold for graph store node matching (#3593)

mem0/graphs/configs.py

@@ -89,6 +89,15 @@ class GraphStoreConfig(BaseModel):
     custom_prompt: Optional[str] = Field(
         description="Custom prompt to fetch entities from the given text", default=None
     )
+    threshold: float = Field(
+        description="Threshold for embedding similarity when matching nodes during graph ingestion. "
+                    "Range: 0.0 to 1.0. Higher values require closer matches. "
+                    "Use lower values (e.g., 0.5-0.7) for distinct entities with similar embeddings. "
+                    "Use higher values (e.g., 0.9+) when you want stricter matching.",
+        default=0.7,
+        ge=0.0,
+        le=1.0,
+    )
 
     @field_validator("config")
     def validate_config(cls, v, values):

mem0/graphs/neptune/base.py

@@ -234,8 +234,8 @@ class NeptuneBase(ABC):
             dest_embedding = self.embedding_model.embed(destination)
 
             # search for the nodes with the closest embeddings
-            source_node_search_result = self._search_source_node(source_embedding, user_id, threshold=0.9)
-            destination_node_search_result = self._search_destination_node(dest_embedding, user_id, threshold=0.9)
+            source_node_search_result = self._search_source_node(source_embedding, user_id, threshold=self.threshold)
+            destination_node_search_result = self._search_destination_node(dest_embedding, user_id, threshold=self.threshold)
 
             cypher, params = self._add_entities_cypher(
                 source_node_search_result,

mem0/graphs/neptune/neptunedb.py

@@ -56,7 +56,8 @@ class MemoryGraph(NeptuneBase):
 
         self.llm = NeptuneBase._create_llm(self.config, self.llm_provider)
         self.user_id = None
-        self.threshold = 0.7
+        # Use threshold from graph_store config, default to 0.7 for backward compatibility
+        self.threshold = self.config.graph_store.threshold if hasattr(self.config.graph_store, 'threshold') else 0.7
         self.vector_store_limit=5
 
     def _delete_entities_cypher(self, source, destination, relationship, user_id):

mem0/graphs/neptune/neptunegraph.py

@@ -39,7 +39,8 @@ class MemoryGraph(NeptuneBase):
 
         self.llm = NeptuneBase._create_llm(self.config, self.llm_provider)
         self.user_id = None
-        self.threshold = 0.7
+        # Use threshold from graph_store config, default to 0.7 for backward compatibility
+        self.threshold = self.config.graph_store.threshold if hasattr(self.config.graph_store, 'threshold') else 0.7
 
     def _delete_entities_cypher(self, source, destination, relationship, user_id):
         """