refactor(llm): enhance a string of graph query method

hugegraph-llm/src/hugegraph_llm/demo/gremlin_generate_web_demo.py

@@ -91,7 +91,7 @@ def llm_settings(llm_type):
                     ]
             else:
                 llm_config_input = []
-            llm_config_button = gr.Button("apply configuration")
+            llm_config_button = gr.Button("Apply Configuration")
 
             def apply_configuration(arg1, arg2, arg3, arg4):
                 llm_option = settings.llm_type
@@ -139,7 +139,7 @@ def embedding_settings(embedding_type):
                     ]
             else:
                 embedding_config_input = []
-            embedding_config_button = gr.Button("apply configuration")
+            embedding_config_button = gr.Button("Apply Configuration")
 
             def apply_configuration(arg1, arg2, arg3):
                 embedding_option = settings.embedding_type

hugegraph-llm/src/hugegraph_llm/demo/rag_demo/configs_block.py

@@ -192,7 +192,7 @@ def create_configs_block():
                 gr.Textbox(value=settings.graph_pwd, label="pwd", type="password"),
                 gr.Textbox(value=settings.graph_space, label="graphspace(Optional)"),
             ]
-        graph_config_button = gr.Button("Apply config")
+        graph_config_button = gr.Button("Apply Configuration")
     graph_config_button.click(apply_graph_config, inputs=graph_config_input)  # pylint: disable=no-member
 
     with gr.Accordion("2. Set up the LLM.", open=False):
@@ -227,7 +227,7 @@ def llm_settings(llm_type):
                     ]
             else:
                 llm_config_input = []
-            llm_config_button = gr.Button("apply configuration")
+            llm_config_button = gr.Button("Apply Configuration")
             llm_config_button.click(apply_llm_config, inputs=llm_config_input)  # pylint: disable=no-member
 
     with gr.Accordion("3. Set up the Embedding.", open=False):
@@ -262,7 +262,7 @@ def embedding_settings(embedding_type):
             else:
                 embedding_config_input = []
 
-            embedding_config_button = gr.Button("apply configuration")
+            embedding_config_button = gr.Button("Apply Configuration")
 
             # Call the separate apply_embedding_configuration function here
             embedding_config_button.click(  # pylint: disable=no-member
@@ -299,7 +299,7 @@ def reranker_settings(reranker_type):
                     ]
             else:
                 reranker_config_input = []
-            reranker_config_button = gr.Button("apply configuration")
+            reranker_config_button = gr.Button("Apply Configuration")
 
             # TODO: use "gr.update()" or other way to update the config in time (refactor the click event)
             # Call the separate apply_reranker_configuration function here

hugegraph-llm/src/hugegraph_llm/indices/vector_index.py

@@ -42,6 +42,7 @@ def from_index_file(dir_path: str) -> "VectorIndex":
         if not os.path.exists(index_file) or not os.path.exists(properties_file):
             log.warning("No index file found, create a new one.")
             return VectorIndex()
+
         faiss_index = faiss.read_index(index_file)
         embed_dim = faiss_index.d
         with open(properties_file, "rb") as f:
@@ -54,6 +55,7 @@ def from_index_file(dir_path: str) -> "VectorIndex":
     def to_index_file(self, dir_path: str):
         if not os.path.exists(dir_path):
             os.makedirs(dir_path)
+
         index_file = os.path.join(dir_path, INDEX_FILE_NAME)
         properties_file = os.path.join(dir_path, PROPERTIES_FILE_NAME)
         faiss.write_index(self.index, index_file)
@@ -63,6 +65,7 @@ def to_index_file(self, dir_path: str):
     def add(self, vectors: List[List[float]], props: List[Any]):
         if len(vectors) == 0:
             return
+
         if self.index.ntotal == 0 and len(vectors[0]) != self.index.d:
             self.index = faiss.IndexFlatL2(len(vectors[0]))
         self.index.add(np.array(vectors))
@@ -73,6 +76,7 @@ def remove(self, props: Union[Set[Any], List[Any]]) -> int:
             props = set(props)
         indices = []
         remove_num = 0
+
         for i, p in enumerate(self.properties):
             if p in props:
                 indices.append(i)
@@ -81,15 +85,20 @@ def remove(self, props: Union[Set[Any], List[Any]]) -> int:
         self.properties = [p for i, p in enumerate(self.properties) if i not in indices]
         return remove_num
 
-    def search(self, query_vector: List[float], top_k: int) -> List[Dict[str, Any]]:
+    def search(self, query_vector: List[float], top_k: int, dis_threshold: float = 0.9) -> List[Dict[str, Any]]:
         if self.index.ntotal == 0:
             return []
+
         if len(query_vector) != self.index.d:
             raise ValueError("Query vector dimension does not match index dimension!")
-        _, indices = self.index.search(np.array([query_vector]), top_k)
+
+        distances, indices = self.index.search(np.array([query_vector]), top_k)
         results = []
-        for i in indices[0]:
-            results.append(deepcopy(self.properties[i]))
+        for dist, i in zip(distances[0], indices[0]):
+            if dist < dis_threshold: # Smaller distances indicate higher similarity
+                results.append(deepcopy(self.properties[i]))
+            else:
+                log.debug("Distance %s is larger than threshold %s, ignore this result.", dist, dis_threshold)
         return results
 
     @staticmethod