Upload app.py

app.py

@@ -9,23 +9,12 @@ import os
 sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'utils'))
 from clean_text import clean_text
 from semantic_similarity import Encoder
-from ranker import compute_bayesian_popularity_score
 from main import get_recommendations
 
 print("Loading restaurant data...")
 data = pd.read_csv("data/toy_data_aggregated_embeddings.csv")
 print(f"Loaded {len(data)} restaurants")
 
-# # Compute Bayesian popularity scores
-# print("Computing popularity scores...")
-# data = compute_bayesian_popularity_score(data)
-# print("Popularity scores computed")
-
-# Load embeddings
-print("Loading pre-computed embeddings...")
-all_desc_embeddings = np.vstack(data["embedding"].values)
-print(f"Loaded embeddings with shape {all_desc_embeddings.shape}")
-
 # Initialize semantic encoder
 print("Loading semantic encoder model...")
 try:
@@ -65,135 +54,9 @@ def create_paris_map(results_df):
     
     return m._repr_html_()
 
-# def semantic_search(query, data_source, num_results, use_popularity):
-#     """Semantic search using embeddings"""
-#     if not query.strip():
-#         return "Please enter a search query", None
-    
-#     try:
-#         query_clean = clean_text(query)
-        
-#         # Generate query embedding
-#         print(f"Encoding query: {query_clean}")
-#         query_embedding = encoder.encode([query_clean], show_progress_bar=False)
-#         query_embedding = query_embedding.cpu().numpy()
-        
-#         # Compute semantic similarity
-#         similarities = cosine_similarity(query_embedding, all_desc_embeddings)[0]
-        
-#         # Combine with popularity if requested
-#         if use_popularity:
-#             sim_normalized = (similarities - similarities.min()) / (similarities.max() - similarities.min() + 1e-10)
-#             pop_normalized = (data["pop_score"] - data["pop_score"].min()) / (data["pop_score"].max() - data["pop_score"].min() + 1e-10)
-#             # Combined score: 70% semantic, 30% popularity
-#             scores = 0.7 * sim_normalized + 0.3 * pop_normalized
-#         else:
-#             scores = similarities
-        
-#         top_indices = np.argsort(scores)[-int(num_results):][::-1]
-#         results = data.iloc[top_indices].copy()
-#         results['similarity_score'] = scores[top_indices]
-        
-#         map_html = create_paris_map(results)
-        
-#         output = f"Found {len(results)} restaurants for '{query}'\n"
-#         output += f"Data Source: {data_source}\n"
-#         output += f"Search Method: Semantic Search {'+ Popularity' if use_popularity else ''}\n\n"
-        
-#         for idx, (_, row) in enumerate(results.iterrows(), 1):
-#             name = row.get('name', 'Unknown')
-#             rating = row.get('overall_rating', 'N/A')
-#             reviews = row.get('review_count', 'N/A')
-#             similarity = row.get('similarity_score', 0)
-#             pop_score = row.get('pop_score', 0)
-            
-#             output += f"{idx}. **{name}**\n"
-#             output += f"   Rating: {rating} | Reviews: {reviews}\n"
-#             output += f"   Match: {similarity:.3f}"
-#             if use_popularity:
-#                 output += f" | Popularity: {pop_score:.2f}"
-#             output += "\n"
-            
-#             if 'address' in row and pd.notna(row['address']):
-#                 addr = str(row['address'])[:100]
-#                 output += f"   Address: {addr}\n"
-            
-#             output += "\n"
-        
-#         return output, map_html
-        
-#     except Exception as e:
-#         import traceback
-#         return f"Error: {str(e)}\n\n{traceback.format_exc()}", None
-
-# def keyword_search(query, data_source, num_results, use_popularity):
-#     """Keyword-based search with optional popularity ranking"""
-#     if not query.strip():
-#         return "Please enter a search query", None
-    
-#     try:
-#         query_clean = clean_text(query).lower()
-#         query_words = set(query_clean.split())
-        
-#         scores = []
-#         for idx, row in data.iterrows():
-#             score = 0
-#             name = str(row.get('name', '')).lower()
-            
-#             # Check name matches
-#             for word in query_words:
-#                 if word in name:
-#                     score += 2
-            
-#             rating = float(row.get('overall_rating', 0))
-#             score += rating * 0.5
-            
-#             # Add popularity if requested
-#             if use_popularity:
-#                 pop_score = float(row.get('pop_score', 0))
-#                 score += pop_score * 0.3
-            
-#             scores.append(score)
-        
-#         top_indices = np.argsort(scores)[-int(num_results):][::-1]
-#         results = data.iloc[top_indices].copy()
-#         results['match_score'] = [scores[i] for i in top_indices]
-        
-#         map_html = create_paris_map(results)
-        
-#         output = f"Found {len(results)} restaurants for '{query}'\n"
-#         output += f"Data Source: {data_source}\n"
-#         output += f"Search Method: Keyword Search {'+ Popularity' if use_popularity else ''}\n\n"
-        
-#         for idx, (_, row) in enumerate(results.iterrows(), 1):
-#             name = row.get('name', 'Unknown')
-#             rating = row.get('overall_rating', 'N/A')
-#             reviews = row.get('review_count', 'N/A')
-#             match = row.get('match_score', 0)
-#             pop_score = row.get('pop_score', 0)
-            
-#             output += f"{idx}. **{name}**\n"
-#             output += f"   Rating: {rating} | Reviews: {reviews}\n"
-#             output += f"   Match Score: {match:.2f}"
-#             if use_popularity:
-#                 output += f" | Popularity: {pop_score:.2f}"
-#             output += "\n"
-            
-#             if 'address' in row and pd.notna(row['address']):
-#                 addr = str(row['address'])[:100]
-#                 output += f"   Address: {addr}\n"
-            
-#             output += "\n"
-        
-#         return output, map_html
-        
-#     except Exception as e:
-#         import traceback
-#         return f"Error: {str(e)}\n\n{traceback.format_exc()}", None
-
 
 def search_restaurants(query_input, data_source, num_results):
-    n_candidates = 100
+    n_candidates = 2000
     query_clean = clean_text(query_input)
     restaurant_ids = get_recommendations(query_clean, n_candidates, num_results, data_source)
     
@@ -250,13 +113,6 @@ with gr.Blocks(
             )
     
     with gr.Row():
-        # with gr.Column(scale=2):
-        #     search_method = gr.Radio(
-        #         choices=["Keyword Search", "Semantic Search"],
-        #         value="Semantic Search" if use_semantic else "Keyword Search",
-        #         label="Search Method",
-        #         info="Semantic uses AI embeddings, Keyword uses exact matches"
-        #     )
         
         with gr.Column(scale=1):
             num_results = gr.Slider(
@@ -266,13 +122,6 @@ with gr.Blocks(
                 step=5,
                 label="Results"
             )
-        
-        # with gr.Column(scale=1):
-        #     use_popularity = gr.Checkbox(
-        #         label="Use Popularity Ranking",
-        #         value=True,
-        #         info="Boost popular restaurants"
-        #     )
     
     search_btn = gr.Button("Search Restaurants", variant="primary", size="lg")
     
@@ -320,7 +169,6 @@ with gr.Blocks(
 if __name__ == "__main__":
     print("\nStarting Advanced Restaurant Finder...")
     print(f"{len(data)} restaurants ready to search")
-    print(f"Popularity Ranking: Enabled")
     print("Opening at http://127.0.0.1:7860\n")
     
     # if run locally