Merge pull request #415 from topoteretes/feature/cog-871-remove-dupli…

…cate-edges-from-the-cognify-graph

fix: Fixes duplicated edges in cognify by limiting the recursion dept…

cognee/api/v1/cognify/cognify_v2.py

@@ -94,7 +94,7 @@ async def run_cognify_pipeline(dataset: Dataset, user: User, graph_model: BaseMo
                 summarization_model = cognee_config.summarization_model,
                 task_config = { "batch_size": 10 }
             ),
-            Task(add_data_points, task_config = { "batch_size": 10 }),
+            Task(add_data_points, only_root = True, task_config = { "batch_size": 10 }),
         ]
 
         pipeline = run_tasks(tasks, data_documents, "cognify_pipeline")

cognee/modules/graph/utils/get_graph_from_model.py

@@ -7,6 +7,7 @@ async def get_graph_from_model(
     added_nodes: dict,
     added_edges: dict,
     visited_properties: dict = None,
+    only_root = False,
     include_root = True,
 ):
     if str(data_point.id) in added_nodes:
@@ -86,7 +87,7 @@ async def get_graph_from_model(
             }))
             added_edges[str(edge_key)] = True
 
-        if str(field_value.id) in added_nodes:
+        if str(field_value.id) in added_nodes or only_root:
             continue
 
         property_nodes, property_edges = await get_graph_from_model(

cognee/tasks/storage/add_data_points.py

@@ -5,7 +5,7 @@
 from .index_data_points import index_data_points
 
 
-async def add_data_points(data_points: list[DataPoint]):
+async def add_data_points(data_points: list[DataPoint], only_root = False):
     nodes = []
     edges = []
 
@@ -19,6 +19,7 @@ async def add_data_points(data_points: list[DataPoint]):
             added_nodes = added_nodes,
             added_edges = added_edges,
             visited_properties = visited_properties,
+            only_root = only_root,
         ) for data_point in data_points
     ])