Refactor[#21] : 새로운 임베딩 모델로 변경

.github/workflows/ci.yml

@@ -26,7 +26,7 @@ jobs:
           echo "DB_USERNAME=testuser" >> .env
           echo "DB_PASSWORD=testpass" >> .env
           echo "DB_NAME=testdb" >> .env
-          echo "W2V_MODEL_PATH=./app/assets/word2vec.model" >> .env
+          echo "W2V_MODEL_PATH=./app/assets/node2vec.model" >> .env
           echo "MONGO_URL=mongodb://localhost:27017" >> .env
           echo "DEV_REDIS_HOST=localhost" >> .env
           echo "DEV_REDIS_PORT=6379" >> .env

.gitignore

@@ -16,4 +16,5 @@ __pycache__/
 *.log
 
 # w2v 모델
-GoogleNews-vectors-negative300.bin.gz
+GoogleNews-vectors-negative300.bin.gz
+node2vec.model

app/models/db_w2v_mapper.py

@@ -6,7 +6,7 @@
     "가족": "Family",
     "코미디": "Comedy",
     "애니메이션": "Animation",
-    "SF": "SF",
+    "SF": "Science Fiction",
     "다큐멘터리": "Documentary",
     "판타지": "Fantasy",
     "드라마": "Drama",
@@ -24,5 +24,13 @@
 }
 
 
-def translate_genre(korean_genre: str) -> str:
-    return _genre_mapping.get(korean_genre, korean_genre)
+def translate_genre(genre: str) -> str | None:
+    genre = genre.strip()
+
+    if genre in _genre_mapping:
+        return _genre_mapping[genre]
+
+    if genre in _genre_mapping.values():
+        return genre
+
+    return genre

app/models/word2vec_model.py

@@ -1,4 +1,4 @@
-from gensim.models import KeyedVectors
+from gensim.models import Word2Vec
 
 
 class Word2VecModel:
@@ -8,15 +8,15 @@ class Word2VecModel:
     def load_model(cls, model_path: str):
         if cls._model is None:
             print("모델 로딩 중...")
-            cls._model = KeyedVectors.load_word2vec_format(model_path, binary=True)
+            cls._model = Word2Vec.load(model_path)
             print("모델 로드 완료!")
         return cls._model
 
     @classmethod
     def get_vector(cls, word: str):
         if cls._model is None:
             raise RuntimeError("모델이 아직 로드되지 않았습니다.")
-        return cls._model.get_vector(word)
+        return cls._model.wv.get_vector(word)
 
     @classmethod
     def similarity(cls, word1: str, word2: str) -> float:

app/services/weight_strategy.py

@@ -1,25 +1,26 @@
 from app.enum.action_type import ActionType
 
-def convert_to_weight(type : ActionType, value : float) -> float :
-  if type == ActionType.CLICK:
-    return 0.2
-  if type == ActionType.LIKE:
-    return 0.1
-  if type == ActionType.WATCH:
-    if 30 <= value <= 50:
-      return 0.1
-    if 50 <= value <= 70:
-      return 0.2
-    if 70 <= value:
-      return 0.3
-    else:
-      return 0
-  if type == ActionType.RATING:
-    if value <= 2:
-      return 0
-    if value == 3:
-      return 0.2
-    if value == 4:
-      return 0.3
-    if value == 5:
-      return 0.4
+
+def convert_to_weight(type: ActionType, value: float) -> float:
+    if type == ActionType.CLICK:
+        return 0.05
+    if type == ActionType.LIKE:
+        return 0.1
+    if type == ActionType.WATCH:
+        if 30 <= value <= 50:
+            return 0.1
+        if 50 <= value <= 70:
+            return 0.2
+        if 70 <= value:
+            return 0.3
+        else:
+            return 0
+    if type == ActionType.RATING:
+        if value <= 2:
+            return 0
+        if value == 3:
+            return 0.2
+        if value == 4:
+            return 0.3
+        if value == 5:
+            return 0.4