Adjusted search algorithm

filtering and promotion are now done only for pre-defined categories

app.py

@@ -49,7 +49,16 @@ def get_bert_embeddings(sentence, model, tokenizer):
     return embeddings
 
 # a function that return top-K best restaurants
-def compute_cos_sim(query):
+def compute_cos_sim(input):
+    query = ""
+    query += input
+
+    # for el in st.session_state.preferences_1:
+    #     query += el
+    # for el in st.session_state.preferences_2:
+    #     query += el
+    
+    st.write("Your query is", query)
     embedded_query = get_bert_embeddings(query, model, tokenizer)
     embedded_query = embedded_query.numpy()
     top_similar = np.array([])
@@ -171,28 +180,26 @@ def promote_places(preferences):
         descr = [word.lower() for word in st.session_state.df['Strings'][i].split()]
         name = st.session_state.df['Names'][i]
         for pref in preferences:
-            if pref in descr:
+            if (pref in descr) & ((pref in st.session_state.food) or (pref in st.session_state.ambiance)):
                 st.session_state.df['Weights'][i] = 2 * st.session_state.df['Weights'][i]
 
     return st.session_state.df
 
 def generate_results(sort_by):
     if sort_by == 'Price':
-        with st.spinner("Sorting your results by price..."):
-            st.write("Sorting your results by price...")
-            results = sort_by_price(10)
+        results = sort_by_price(10)
     elif sort_by == 'Rating':
-        with st.spinner("Sorting your results by rating..."):
-            st.write("Sorting your results by rating...")
-            results = sort_by_rating(10)
+        # with st.spinner("Sorting your results by rating..."):
+        #     st.write("Sorting your results by rating...")
+        results = sort_by_rating(10)
     elif sort_by == 'Relevancy (default)':
-        with st.spinner("Sorting your results by relevancy..."):
-            st.write("Sorting your results by relevancy...")
-            results = sort_by_relevancy(10)
+        # with st.spinner("Sorting your results by relevancy..."):
+        #     st.write("Sorting your results by relevancy...")
+        results = sort_by_relevancy(10)
     else:
-        st.write("Sorry, we are still working on this option. For now, the results are sorted by relevance")
-        with st.spinner("Sorting your results by relevancy..."):
-            results = sort_by_relevancy(10)
+        st.write(":pensive: Sorry, we are still working on this option. For now, the results are sorted by relevance")
+        # with st.spinner("Sorting your results by relevancy..."):
+        results = sort_by_relevancy(10)
     return results
 
 if 'preferences_1' not in st.session_state:
@@ -201,6 +208,12 @@ if 'preferences_1' not in st.session_state:
 if 'preferences_2' not in st.session_state:
   st.session_state.preferences_2 = []
   
+if 'additional_1' not in st.session_state:
+  st.session_state.additional_1 = []
+  
+if 'additional_2' not in st.session_state:
+  st.session_state.additional_2 = []
+  
 if 'food' not in st.session_state:
   st.session_state.food = ['Coffee', 'Italian', 'Mexican', 'Chinese', 'Indian', 'Asian', 'Fast food', 'Other']
   
@@ -224,9 +237,6 @@ if 'df' not in st.session_state:
     
 if 'precalculated_df' not in st.session_state:
     st.session_state.precalculated_df = pd.DataFrame()
-    
-if 'stop_search' not in st.session_state:
-    st.session_state.stop_search = False
 
 # Configure Streamlit page and state
 st.title("GoTogether!")
@@ -299,7 +309,7 @@ if food_1 == 'Other':
 ambiance_1 = st.selectbox('What describes your occasion the best?', st.session_state.ambiance, key=2)
 if ambiance_1 == 'Other':
     ambiance_1 = st.text_input(label="Your description", placeholder="How would you describe your meeting?", key=11)
-
+    
 options_food_1 = st.multiselect(
     'Do you have any dietary restrictions?',
     ['Vegan', 'Vegetarian', 'Halal'], key=100)
@@ -329,16 +339,33 @@ with_kids_2 = st.checkbox('I will come with kids', key=201)
     
 if len(st.session_state.preferences_1) == 0:
     st.session_state.preferences_1.append(food_1)
+    # if food_1 in st.session_state.food:
+    #     st.session_state.preferences_1.append(food_1)
+    # else:
+    #     st.session_state.additional_1.append(food_1_o)
     st.session_state.preferences_1.append(ambiance_1)
+        
+    # if ambiance_1 in st.session_state.ambiance:
+    #     st.session_state.preferences_1.append(ambiance_1)
+    # else:
+    #     st.session_state.additional_1.append(ambiance_1_o)
     st.session_state.restrictions.extend(options_food_1)
-    if additional_1:
-        st.session_state.preferences_1.append(additional_1)
     if with_kids:
         st.session_state.restrictions.append('kids')
-
+    if additional_1:
+        st.session_state.preferences_1.append(additional_1)
+    
 if len(st.session_state.preferences_2) == 0:
     st.session_state.preferences_2.append(food_2)
+    # if food_2 in st.session_state.food:
+    #     st.session_state.preferences_2.append(food_2)
+    # else:
+    #     st.session_state.additional_2.append(food_2_o)
     st.session_state.preferences_2.append(ambiance_2)
+    # if ambiance_2 in st.session_state.ambiance:
+    #     st.session_state.preferences_2.append(ambiance_2)
+    # else:
+    #     st.session_state.additional_2.append(ambiance_2_o)
     st.session_state.restrictions.extend(options_food_2)
     if additional_2:
         st.session_state.preferences_2.append(additional_2)
@@ -348,8 +375,9 @@ if len(st.session_state.preferences_2) == 0:
 submitted = st.button('Submit!')
     
 if submitted:
-    st.markdown("Thanks, we received your preferences!")
-    st.session_state.stop_search = False
+    with st.spinner('Processing your request...'):
+        time.sleep(1)
+    st.success("Thanks, we received your preferences!")
     
 else:
     st.write('☝️ Describe your preferences!')
@@ -371,48 +399,51 @@ if submit or (not st.session_state.precalculated_df.empty):
                                index=st.session_state.options.index('Relevancy (default)'))
         if sort_by:
             st.session_state.sort_by = sort_by
-            results = generate_results(st.session_state.sort_by)
-            k = 10
-            st.write(f"Here are the best {k} matches to your preferences:")
-            i = 1
-            nums = list(range(1, 11))
-            words = ['one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', 'nine', 'one: :zero']
-            nums_emojis = dict(zip(nums, words))
-            for name, score in results.items():
-                
-                condition = st.session_state.precalculated_df['Names'] == name
-                rating = st.session_state.precalculated_df.loc[condition, 'Rating'].values[0]
-                with st.expander(f":{nums_emojis[i]}: **{name}** **({str(rating)}**:star:): match score: {score}"):
+            with st.spinner(f"Sorting your results by {sort_by.lower()}..."):
+                results = generate_results(st.session_state.sort_by)
+                k = 10
+                st.write(f"Here are the best {k} matches to your preferences:")
+                i = 1
+                nums = list(range(1, 11))
+                words = ['one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', 'nine', 'one: :zero']
+                nums_emojis = dict(zip(nums, words))
+                for name, score in results.items():
                     
-                    #f":{nums_emojis[i]}: **{name}** **({str(rating)}**:star:) :", 'match score:', score
-                    try:
-                        if type(st.session_state.precalculated_df.loc[condition, 'Price'].values[0]) == str:
-                            st.write("Price category:", st.session_state.precalculated_df.loc[condition, 'Price'].values[0])
-                    except:
-                        pass
+                    condition = st.session_state.precalculated_df['Names'] == name
+                    rating = st.session_state.precalculated_df.loc[condition, 'Rating'].values[0]
+                    with st.expander(f":{nums_emojis[i]}: **{name}** **({str(rating)}**:star:): match score: {score}"):
                         
-                    # Use the condition to extract the value(s)
-                    # description = st.session_state.precalculated_df.loc[condition, 'Strings']
-                    # st.write(description)
-                    
-                    type = [item for item in eval(st.session_state.precalculated_df.loc[condition, 'Category'].values[0])]
-                    # Display HTML with the custom styles
-                    for word in type:
-                        st.markdown(css, unsafe_allow_html=True)
-                        st.markdown(f'<div class="blue-box">{word}</div>', unsafe_allow_html=True)
-                    # st.write("Restaurant type:", str(type))
-                    
-                    keywords = [item[0] for item in eval(st.session_state.precalculated_df.loc[condition, 'Keywords'].values[0])]
-                    for pair in keywords[:3]:
-                        st.markdown(css, unsafe_allow_html=True)
-                        st.markdown(f'<div class="orange-box">{pair[0]} {pair[1]}</div>', unsafe_allow_html=True)
-                    # st.write("Restaurant type:", str(type))
-                    
-                    
-                    url = st.session_state.precalculated_df.loc[condition, 'URL'].values[0]
-                    st.write(f"_Check on the_ [_map_]({url})")
-                                    
-                i+=1
+                        #f":{nums_emojis[i]}: **{name}** **({str(rating)}**:star:) :", 'match score:', score
+                        try:
+                            if type(st.session_state.precalculated_df.loc[condition, 'Price'].values[0]) == str:
+                                st.write("Price category:", st.session_state.precalculated_df.loc[condition, 'Price'].values[0])
+                        except:
+                            pass
+                            
+                        # Use the condition to extract the value(s)
+                        # description = st.session_state.precalculated_df.loc[condition, 'Strings']
+                        # st.write(description)
+                        
+                        type = [item for item in eval(st.session_state.precalculated_df.loc[condition, 'Category'].values[0])]
+                        # Display HTML with the custom styles
+                        for word in type:
+                            st.markdown(css, unsafe_allow_html=True)
+                            st.markdown(f'<div class="blue-box">{word}</div>', unsafe_allow_html=True)
+                        # st.write("Restaurant type:", str(type))
+                        
+                        keywords = [item[0] for item in eval(st.session_state.precalculated_df.loc[condition, 'Keywords'].values[0]) if item[1] > 2]
+                        for pair in keywords[:3]:
+                            st.markdown(css, unsafe_allow_html=True)
+                            st.markdown(f'<div class="orange-box">{pair[0]} {pair[1]}</div>', unsafe_allow_html=True)
+                        # st.write("Restaurant type:", str(type))
+                        
+                        
+                        url = st.session_state.precalculated_df.loc[condition, 'URL'].values[0]
+                        st.write(f"_Check on the_ [_map_]({url})")
+                        
+                        st.write(st.session_state.precalculated_df.loc[condition, 'Strings'].values[0])
+                                        
+                    i+=1
                             
                 # st.markdown("This is a text with <span style='font-size: 20px;'>bigger</span> and <i>italic</i> text.", unsafe_allow_html=True)
                 # st.markdown("<span style='font-size: 24px;'>This is larger text</span>", unsafe_allow_html=True)
@@ -425,6 +456,7 @@ if stop:
     st.write("New search is launched. Please specify your preferences in the form!")
     st.session_state.preferences_1, st.session_state.preferences_2 = [], []
     st.session_state.restrictions = []
+    st.session_state.additional_1, st.session_state.additional_2 = [], []
     st.session_state.sort_by = ""
     st.session_state.df = init_df
     st.session_state.precalculated_df = pd.DataFrame()