Update predictor.py

predictor.py

@@ -1,6 +1,6 @@
 import streamlit as st
 import pandas as pd
-import re 
+import re
 from utilities import predict
 
 st.set_page_config(layout="wide")
@@ -67,39 +67,33 @@ def create_expander_with_check_button(label, title, context, predict_func):
                     st.warning(f"The predicted probability for label '{label}' is too high or the predicted label is different. Please modify the claim.")
         else:
             st.warning("Please enter a claim.")
-
-
+            
 if 'annotated_data' not in st.session_state:
-    st.session_state['annotated_data'] = pd.DataFrame(columns=['Context ID', 'Context', 'Claim', 'Label', 'Evidence', 'Title', 'Link', 'Claim ID'])
+    st.session_state['annotated_data'] = pd.DataFrame(columns=['Context', 'Claim', 'Label', 'Evidence', 'Title', 'Link'])
 
 annotated_data = st.session_state['annotated_data']
 
-def save_data(context, default_title, default_link, context_id):
+def save_data(context, default_title, default_link):
     annotated_data = st.session_state['annotated_data']
     error = 'success'
     
-    for label, label_index in [('SUPPORTED', 0), ('REFUTED', 1), ('NEI', 2)]:
+    for label in ['NEI', 'REFUTED', 'SUPPORTED']:
         claim_key = f"{label}_input"
         evidence_key = f"{label}_evidence_selected"
         
         if st.session_state.get(claim_key, ''):
             claim = st.session_state[claim_key]
             evidence = st.session_state.get(evidence_key, [])
-            
-            # Calculate the length of evidence
-            evidence_length = len(evidence) if label != 'NEI' else 0
-            
-            if not annotated_data[((annotated_data['Label'] == label) & (annotated_data['Title'] == default_title))].empty:
+            if label == 'NEI':
+                evidence = []  # No evidence required for NEI
+            if not annotated_data[((annotated_data['Claim'] == claim) & (annotated_data['Label'] == label) & (annotated_data['Title'] == default_title))].empty:
                 error = 'duplicate'
             else:
-                # Create a unique claim ID using context ID, label index, and evidence length
-                claim_id = f"{context_id}_{label_index}_{evidence_length}"
-                # Save the data including the context_id
-                annotated_data.loc[len(annotated_data)] = [context_id, context, claim, label, evidence, default_title, default_link, claim_id]
+                annotated_data.loc[len(annotated_data)] = [context, claim, label, evidence, default_title, default_link]
     
     st.session_state['annotated_data'] = annotated_data
     return error
-    
+
 def enough_claims_entered(title):
     annotated_data = st.session_state['annotated_data']
     nei_claims = annotated_data[(annotated_data['Label'] == 'NEI') & (annotated_data['Title'] == title)].shape[0]
@@ -232,7 +226,7 @@ def predictor_app():
                         save_button = st.button("Save")
                         if save_button:
                             if all_claims_entered and all_evidence_selected:
-                                error = save_data(default_context, default_title, default_link, default_ID)
+                                error = save_data(default_context, default_title, default_link)
                             else:
                                 error = 'save_fail'