251029更新

推荐系统/main.py

@@ -29,7 +29,7 @@ class InitializationArguments(BaseModel):
 
     # 时间窗口（单位为天），平衡实时性和运算效率
     time_window: int = Field(default=30, ge=5, le=360)
-    # 衰减因子兰布达系数，控制兴趣分数衰减速率
+    # 衰减因子兰布达系数，控制兴趣分数衰减速率（默认不衰减）
     decay_lambda: float = Field(default=0, ge=0.00, le=10)
     # 用户特征向量维度数
     attributes_dimensions: int = Field(default=10, ge=2.00, le=200)
@@ -210,9 +210,9 @@ class RecommenderSystem:
         # 根据行为类型获取兴趣基础分数和衰减权重
         score_base, weight = self.behavior_arguments[type_]
 
-        # 若行为类型为评分则将基础分数转化为0.2～0.8
+        # 若行为类型为评分则将评分转为基础分数（基于最小值-最大值归一化为0.2～1.0）
         if type_ == "rating":
-            score_base = 0.1 + 0.8 * (1 / (1 + numpy.exp(3 - rating)))
+            score_base = 0.2 * (rating - 1) + 0.2
 
         return score_base * numpy.exp(0 - time_interval * (self.decay_lambda * weight))
 
@@ -259,8 +259,6 @@ class RecommenderSystem:
         # 基于物品的协同过滤生成推荐物品标识列表
         candidates_items = self._generate_items_candidates(user=user, k=k)
 
-        print(candidates_items)
-
         # 基于用户的协同过滤生成推荐物品标识列表
         candidates_users = self._generate_users_candidates(user=user, k=k)
 
@@ -327,18 +325,17 @@ class RecommenderSystem:
                     else 0
                 )
 
-                # 流行度抑制因子
-                popularity_suppressed = len(
-                    list(set(users_heuristic) & set(users_recall))
-                ) / numpy.sqrt(len(users_heuristic) * len(users_recall))
+                print(pair, similarity)
 
                 # 加权物品的相似度
                 items_recall[item_recall]["scores"] += (
-                    behaviors["scores"][item_heuristic]
-                    * similarity
-                    * popularity_suppressed
+                    behaviors["scores"][item_heuristic] * similarity
                 )
 
+        print(items_recall)
+
+        exit()
+
         return self._normalize_scores(items_recall=items_recall, k=k)
 
     # 基于用户协同过滤算法生成候选物品标识列表
@@ -385,8 +382,6 @@ class RecommenderSystem:
         # 候选物品标识列表
         candidates = defaultdict(float)
 
-        print(items_recall)
-
         if items_recall:
             scores = [value["scores"] for value in items_recall.values()]
 
@@ -428,7 +423,7 @@ if __name__ == "__main__":
             "user": "aaaaaa",
             "item": "111111",
             "type_": "rating",
-            "timestamp": int(time.time() - 3600),
+            "timestamp": int(time.time() - 3200),
             "rating": 4,
         },
         {