[add] The application layer introduces the clustering community-retrieval module

api/app/repositories/neo4j/create_indexes.py

@@ -42,6 +42,13 @@ async def create_fulltext_indexes():
             OPTIONS { indexConfig: { `fulltext.analyzer`: 'cjk' } }
         """)
         print("✓ Created: summariesFulltext")
+
+        # 创建 Community 索引
+        await connector.execute_query("""
+            CREATE FULLTEXT INDEX communitiesFulltext IF NOT EXISTS FOR (c:Community) ON EACH [c.name, c.summary]
+            OPTIONS { indexConfig: { `fulltext.analyzer`: 'cjk' } }
+        """)
+        print("✓ Created: communitiesFulltext")
         
         print("\nFull-text indexes created successfully with BM25 support.")
     except Exception as e:
@@ -112,6 +119,18 @@ async def create_vector_indexes():
             }}
         """)
         print("✓ Created: summary_embedding_index")
+
+        # Community summary embedding index
+        await connector.execute_query("""
+            CREATE VECTOR INDEX community_summary_embedding_index IF NOT EXISTS
+            FOR (c:Community)
+            ON c.summary_embedding
+            OPTIONS {indexConfig: {
+              `vector.dimensions`: 1024,
+              `vector.similarity_function`: 'cosine'
+            }}
+        """)
+        print("✓ Created: community_summary_embedding_index")
         
         # Dialogue embedding index (optional)
         await connector.execute_query("""

api/app/repositories/neo4j/graph_search.py

@@ -305,6 +305,7 @@ async def search_graph(
     results = {}
     for key, result in zip(task_keys, task_results):
         if isinstance(result, Exception):
+            logger.warning(f"search_graph: {key} 关键词查询异常: {result}")
             results[key] = []
         else:
             results[key] = result
@@ -361,7 +362,11 @@ async def search_graph_by_embedding(
     print(f"[PERF] Embedding generation took: {embed_time:.4f}s")
     
     if not embeddings or not embeddings[0]:
-        return {"statements": [], "chunks": [], "entities": [], "summaries": []}
+        logger.warning(
+            f"search_graph_by_embedding: embedding 生成失败或为空，"
+            f"query='{query_text[:50]}', end_user_id={end_user_id}，向量检索跳过"
+        )
+        return {"statements": [], "chunks": [], "entities": [], "summaries": [], "communities": []}
     embedding = embeddings[0]
 
     # Prepare tasks for parallel execution
@@ -435,6 +440,7 @@ async def search_graph_by_embedding(
     
     for key, result in zip(task_keys, task_results):
         if isinstance(result, Exception):
+            logger.warning(f"search_graph_by_embedding: {key} 向量查询异常: {result}")
             results[key] = []
         else:
             results[key] = result