feat: Model updated | 53번 채점을 LSA로 적용

Model/app.py

@@ -6,6 +6,7 @@
 import json
 import re
 from gpt_response import gpt_response
+from lsa_Similar import lsa_Similar
 from flask_cors import CORS
 app = Flask(__name__) #상태 알아보기2
 cors = CORS(app, resources={r"*": {"origins": "https://port-0-docker-essay-score-jvvy2blm7ipnj3.sel5.cloudtype.app"}})
@@ -279,7 +280,10 @@ def get_score():
        return jsonify(response)
     else:
       #사용자 답안과 , 실제 답안 content, answer
-      similar = similarity(contents, answer)
+      if quest_num<=52:
+        similar = similarity(contents, answer)
+      else:
+         similar = lsa_Similar(contents,answer)
       #사용자 답안 content
       spell = pusan_univ_spell(contents)
 

Model/lsa_Similar.py

@@ -0,0 +1,40 @@
+import numpy as np
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.decomposition import TruncatedSVD
+from sklearn.metrics.pairwise import cosine_similarity
+from PyKomoran import * #형태소 분석기 변경
+
+tfidf_vectorizer = TfidfVectorizer()
+komoran = Komoran('STABLE')
+
+def lsa_Similar(contents, answer):
+    test = (komoran.get_plain_text(contents[0])).split(' ')
+    for j in range(len(test)):
+        temp = test[j].split('/')
+        test[j] = temp[0]
+        #print('여기',test)
+    test = ' '.join(test)
+    test2 = (komoran.get_plain_text(answer[0])).split(' ')
+    for j in range(len(test2)):
+        temp = test2[j].split('/')
+        test2[j] = temp[0]
+        #print('여기',test)
+    test2 = ' '.join(test2)
+    tfidf_vectorizer = TfidfVectorizer()
+    tfidf_matrix = tfidf_vectorizer.fit_transform([test, test2])
+
+    # LSA를 사용하여 차원 축소
+    lsa = TruncatedSVD(n_components=2)
+    lsa_matrix = lsa.fit_transform(tfidf_matrix)
+
+    # 문장 간 유사도 계산
+    similarity_matrix = cosine_similarity(lsa_matrix)
+
+    response = {
+        'best_i': 0,
+        'best_dist': 1 - similarity_matrix[1][0],
+        'result': contents
+    }
+
+    print(similarity_matrix)
+    return response