diff --git a/Merge/back_rhj/app/services/__pycache__/lightgcn_scorer.cpython-312.pyc b/Merge/back_rhj/app/services/__pycache__/lightgcn_scorer.cpython-312.pyc
index bd899a9..5e2f64b 100644
--- a/Merge/back_rhj/app/services/__pycache__/lightgcn_scorer.cpython-312.pyc
+++ b/Merge/back_rhj/app/services/__pycache__/lightgcn_scorer.cpython-312.pyc
Binary files differ
diff --git a/Merge/back_rhj/app/services/__pycache__/recommendation_service.cpython-312.pyc b/Merge/back_rhj/app/services/__pycache__/recommendation_service.cpython-312.pyc
index 0446ed2..8eac212 100644
--- a/Merge/back_rhj/app/services/__pycache__/recommendation_service.cpython-312.pyc
+++ b/Merge/back_rhj/app/services/__pycache__/recommendation_service.cpython-312.pyc
Binary files differ
diff --git a/Merge/back_rhj/app/services/lightgcn_scorer.py b/Merge/back_rhj/app/services/lightgcn_scorer.py
index f6aeb19..30cddd4 100644
--- a/Merge/back_rhj/app/services/lightgcn_scorer.py
+++ b/Merge/back_rhj/app/services/lightgcn_scorer.py
@@ -121,6 +121,8 @@
             List[float]: 每个候选物品的LightGCN分数
         """
         self._initialize_model()
+        print(self.user2idx, "用户映射")
+        print(self.post2idx, "物品映射")
         
         # 检查用户是否存在
         if user_id not in self.user2idx:
diff --git a/Merge/back_rhj/app/services/recommendation_service.py b/Merge/back_rhj/app/services/recommendation_service.py
index 0547f7b..8982bd3 100644
--- a/Merge/back_rhj/app/services/recommendation_service.py
+++ b/Merge/back_rhj/app/services/recommendation_service.py
@@ -27,7 +27,7 @@
         args.data_path = './app/user_post_graph.txt'  # 修改为帖子图文件
         args.pre_model_path = './app/models/recommend/LightGCN_pretrained.pt'
         
-        self.topk = 2  # 默认推荐数量
+        self.topk = 10  # 默认推荐数量
         
         # 初始化多路召回管理器
         self.multi_recall = None
@@ -82,7 +82,7 @@
         
         return intersection / union if union > 0 else 0.0
     
-    def mmr_rerank_with_ads(self, post_ids, scores, theta=0.5, target_size=None):
+    def mmr_rerank_with_ads(self, post_ids, scores, theta=0.7, target_size=None):
         """
         使用MMR算法重新排序推荐结果，并在过程中加入广告约束
         输入: