Update main.py

main.py

@@ -152,18 +152,55 @@ def parse_news(req: NewsRequest):
 
  
         resultK = resultKeyword(content)
-        sumce = classify_emotion(content)
         targetCompany = gemini_use(resultK)
 
 
         sentiment = analyze_sentiment(content)
-        pos_percent = int(sentiment["positive"] * 100)
-        neg_percent = int(sentiment["negative"] * 100)
-
-        sentiment_result = {
-            "positive": pos_percent,
-            "negative": neg_percent
-        }
+        pos_score = sentiment["positive"]
+        neg_score = sentiment["negative"]
+        net_score = sentiment["neutral"]
+        print("부정:", sentiment["negative"])
+        print("중립:", sentiment["neutral"])
+        print("긍정:", sentiment["positive"])
+
+        # 중립 점수 절반으로 줄이고 나머지를 부정/긍정에 재분배
+        reduced_net = net_score / 2
+        remaining = net_score - reduced_net
+        total_non_neu = neg_score + pos_score
+
+        if total_non_neu > 0:
+            neg_score += remaining * (neg_score / total_non_neu)
+            pos_score += remaining * (pos_score / total_non_neu)
+        else:
+            neg_score += remaining / 2
+            pos_score += remaining / 2
+
+        net_score = reduced_net
+
+        max_label = max(
+            [("부정", neg_score), ("중립", net_score), ("긍정", pos_score)],
+            key=lambda x: x[1]
+        )[0]
+
+        if max_label == "긍정":
+            if pos_score >= 0.9:
+                sentiment_label = f"매우 긍정 ({pos_score*100:.1f}%)"
+            elif pos_score >= 0.6:
+                sentiment_label = f"긍정 ({pos_score*100:.1f}%)"
+            else:
+                sentiment_label = f"약한 긍정 ({pos_score*100:.1f}%)"
+        elif max_label == "부정":
+            if neg_score >= 0.9:
+                sentiment_label = f"매우 부정 ({neg_score*100:.1f}%)"
+            elif neg_score >= 0.6:
+                sentiment_label = f"부정 ({neg_score*100:.1f}%)"
+            else:
+                sentiment_label = f"약한 부정 ({neg_score*100:.1f}%)"
+        else:
+            sentiment_label = f"중립 ({net_score*100:.1f}%)"
+                
+#밑에 부분은 모델로 주가 예측 한거임
+    
 
         summary = summarize(content)
         print(summary)
@@ -201,8 +238,8 @@ def parse_news(req: NewsRequest):
             "summary": resultK["summary"],
             "keyword": resultK["keyword"],
             "company": targetCompany,
-            "sentiment": sumce,
-            "sentiment_value": sentiment_result,
+            "sentiment": sentiment_label,
+            "sentiment_value": sentiment_label,
             "prediction": prediction,
             "prob": prob,
         }