Update app.py

app.py

@@ -69,14 +69,8 @@ def filter_semantically_similar_texts_by_embedding(df, embedding_field='embeddin
 
     return df.iloc[filtered_indices].reset_index(drop=True)
     
-def find_ori_num(list_number):
-    if not (5 <= list_number <= 15):
-        return "5~15사이의 값을 입력해주세요"
-    if list_number:
-        ori_num = 15 - list_number
-        return ori_num
-
-def search_kpi(kpi_query, val=None):
+
+def search_kpi(kpi_query, kpi_count):
     results = kpi_pool.similarity_search_with_relevance_scores(kpi_query, k=50)
 
     # 메타데이터 + 점수 추출
@@ -85,16 +79,11 @@ def search_kpi(kpi_query, val=None):
         for doc, score in results
     ]
 
+    # DataFrame으로 변환
     df = pd.DataFrame(records)
-    df['embedding'] = df['지표명']+" "+df['정의']
-
-    # 다양성 필터링 (텍스트 임베딩 기반)    
-    min_num = find_ori_num(val)
-    if val:
-        df = filter_semantically_similar_texts_by_embedding(df, embedding_field='embedding', similarity_threshold=0.8)
-        df = df.iloc[min_num:15]
-    else:
-        df = filter_semantically_similar_texts_by_embedding(df, embedding_field='embedding', similarity_threshold=0.8).head(15)
+    df['카테고리'] = df['BSC 관점'] + " > " + df['전략방향']
+    df = df.drop_duplicates(subset=['정의', '산식']).head(15)
+    df = df.iloc[:kpi_count]
 
     # 카테고리 생성 (BSC 관점 + 전략방향)
     df['카테고리'] = df['BSC 관점'] + " > " + df['전략방향']
@@ -238,7 +227,7 @@ with gr.Blocks(css=css) as demo:
                 kpi_query = gr.Textbox(scale=30, submit_btn=True,
                                     label= "성과평가를 진행할 [핵심업무 or 핵심성공요인]을 입력해주세요😊!  (검색 키워드는 직무기술서 또는 NCS 능력단위 참고)",
                                     placeholder="예: 자금계획수립 정확성, 자금조달 적시성, 재무위험관리 효과성 등")
-                val_num = gr.Slider(label="KPI 출력 개수", value = 10, minimum=5, maximum=15, step=1, scale=7)
+                kpi_count = gr.Slider(label="KPI 출력 개수", value = 15, minimum=5, maximum=15, step=1, scale=7)
                 
             copyright = gr.Markdown(guide, visible=True, elem_classes="custom-markdown")
 
@@ -260,7 +249,7 @@ with gr.Blocks(css=css) as demo:
 
             file_download = gr.Files(label="Download", interactive=False, visible=False)
         
-        kpi_query.submit(search_kpi, inputs = [kpi_query,val_num], outputs = [output_area, kpi_checkbox, kpi_table, df_state, check_state, copyright])
+        kpi_query.submit(search_kpi, inputs = [kpi_query, kpi_count], outputs = [output_area, kpi_checkbox, kpi_table, df_state, check_state, copyright])
         select_button.click(fn=toggle_selection, inputs=[kpi_checkbox, check_state], outputs=kpi_checkbox, show_progress='hidden')
         download_button.click(
             generate_excel,