A clear commented version and pipeline summeration in my_progress.md

hirag/_op.py

@@ -140,6 +140,13 @@ async def _handle_entity_relation_summary(
     description: str,
     global_config: dict,
 ) -> str:
+    """Summarize the entity or relation description,is used during entity extraction and when merging nodes or edges in the knowledge graph
+
+    Args:
+        entity_or_relation_name: entity or relation name
+        description: description
+        global_config: global configuration
+    """
     use_llm_func: callable = global_config["cheap_model_func"]
     llm_max_tokens = global_config["cheap_model_max_token_size"]
     tiktoken_model_name = global_config["tiktoken_model_name"]
@@ -311,6 +318,17 @@ async def extract_hierarchical_entities(
     entity_vdb: BaseVectorStorage,
     global_config: dict,
 )-> Union[BaseGraphStorage, None]:
+    """Extract entities and relations from text chunks
+
+    Args:
+        chunks: text chunks
+        knowledge_graph_inst: knowledge graph instance
+        entity_vdb: entity vector database
+        global_config: global configuration
+
+    Returns:
+        Union[BaseGraphStorage, None]: knowledge graph instance
+    """
     use_llm_func: callable = global_config["best_model_func"]
     entity_extract_max_gleaning = global_config["entity_extract_max_gleaning"]
 

hirag/_storage/gdb_networkx.py

@@ -199,7 +199,10 @@ class NetworkXStorage(BaseGraphStorage):
 
     async def _leiden_clustering(self):
         from graspologic.partition import hierarchical_leiden
-
+        """
+        It uses the hierarchical_leiden function from the graspologic library
+        The Leiden algorithm is used in the HiRAG.ainsert method
+        """
         graph = NetworkXStorage.stable_largest_connected_component(self._graph)
         community_mapping = hierarchical_leiden(
             graph,

hirag/prompt.py

@@ -533,7 +533,7 @@ Entity description list: {entity_description_list}
 #######
 Output:
 """
-
+# TYPE的定义
 PROMPTS["DEFAULT_ENTITY_TYPES"] = ["organization", "person", "geo", "event"]
 PROMPTS["META_ENTITY_TYPES"] = ["organization", "person", "location", "event", "product", "technology", "industry", "mathematics", "social sciences"]
 PROMPTS["DEFAULT_TUPLE_DELIMITER"] = "<|>"