Rename app_11.py to app_12.py

app_11.py → app_12.py

@@ -154,11 +154,37 @@ def extract_arguments(text, tokenizer, model, beam_search=True):
     list1 = [start_cause1, end_cause1, start_effect1, end_effect1, start_signal, end_signal]
     list2 = [start_cause2, end_cause2, start_effect2, end_effect2, start_signal, end_signal]
     #return cause1, cause2, effect1, effect2, signal, list1, list2
-    return start_cause1, end_cause1, start_cause2, end_cause2, start_effect1, end_effect1, start_effect2, end_effect2, start_signal, end_signal
+    #return start_cause1, end_cause1, start_cause2, end_cause2, start_effect1, end_effect1, start_effect2, end_effect2, start_signal, end_signal
+
+    
+
+    # Add the argument tags in the sentence directly
+    def add_tags(tokens, start_cause1, end_cause1, start_effect1, end_effect1, start_signal, end_signal):
+        if start_cause is not None and end_cause is not None:
+            tokens[start_cause1] = '<ARG0>' + tokens[start_cause1]
+            tokens[end_cause1] = tokens[end_cause1] + '</ARG0>'
+        
+        if start_effect is not None and end_effect is not None:
+            tokens[start_effect1] = '<ARG1>' + tokens[start_effect1]
+            tokens[end_effect1] = tokens[end_effect1] + '</ARG1>'
+
+        if start_signal is not None and end_signal is not None:
+            tokens[start_signal] = '<SIG0>' + tokens[start_signal]
+            tokens[end_signal] = tokens[end_signal] + '</SIG0>'
+
+        return ' '.join(tokens)
+
+    # Apply the tags to the sentence tokens
+    tagged_sentence1 = add_tags(tokens, start_cause1, end_cause1, start_effect1, end_effect1, start_signal, end_signal)
+    tagged_sentence2 = add_tags(tokens, start_cause2, end_cause2, start_effect2, end_effect2, start_signal, end_signal)
+    return tagged_sentence1, tagged_sentence2
+
+
+
 
 def mark_text_by_position(original_text, start_idx, end_idx, color):
     """Marks text in the original string based on character positions."""
-    if start_idx is not None and end_idx is not None and start_idx < end_idx:
+    if start_idx is not None and end_idx is not None and start_idx <= end_idx:
         return (
             original_text[:start_idx]
             + f"<mark style='background-color:{color}; padding:2px; border-radius:4px;'>"
@@ -171,16 +197,55 @@ def mark_text_by_position(original_text, start_idx, end_idx, color):
 def mark_text_by_tokens(tokenizer, tokens, start_idx, end_idx, color):
     """Highlights a span in tokenized text using HTML."""
     highlighted_tokens = copy.deepcopy(tokens)  # Avoid modifying original tokens
-    if start_idx is not None and end_idx is not None and start_idx < end_idx:
+    if start_idx is not None and end_idx is not None and start_idx <= end_idx:
         highlighted_tokens[start_idx] = f"<span style='background-color:{color}; padding:2px; border-radius:4px;'>{highlighted_tokens[start_idx]}"
         highlighted_tokens[end_idx] = f"{highlighted_tokens[end_idx]}</span>"
     return tokenizer.convert_tokens_to_string(highlighted_tokens)
 
+def mark_text_by_word_ids(original_text, token_ids, start_word_id, end_word_id, color):
+    """Marks words in the original text based on word IDs from tokenized input."""
+    words = original_text.split()  # Split text into words
+    if start_word_id is not None and end_word_id is not None and start_word_id <= end_word_id:
+        words[start_word_id] = f"<mark style='background-color:{color}; padding:2px; border-radius:4px;'>{words[start_word_id]}"
+        words[end_word_id] = f"{words[end_word_id]}</mark>"
+    
+    return " ".join(words)
+
+
+
 
 st.title("Causal Relation Extraction")
 input_text = st.text_area("Enter your text here:", height=300)
 beam_search = st.radio("Enable Beam Search?", ('No', 'Yes')) == 'Yes'
 
+if st.button("Add Argument Tags"):
+    if text_input:
+        tagged_sentence1, tagged_sentence2 = extract_arguments(input_text, tokenizer, model, beam_search=True)
+
+        st.write("**Tagged Sentence_1:**")
+        st.write(tagged_sentence1)
+        st.write("**Tagged Sentence_2:**")
+        st.write(tagged_sentence2)
+    else:
+        st.warning("Please enter some text to analyze.")
+    
+
+if st.button("Extract"):
+    if input_text:
+        start_cause_id, end_cause_id, start_effect_id, end_effect_id, start_signal_id, end_signal_id = extract_arguments(input_text, tokenizer, model, beam_search=beam_search)
+
+        cause_text = mark_text_by_word_ids(input_text, inputs["input_ids"][0], start_cause_id, end_cause_id, "#FFD700")  # Gold for cause
+        effect_text = mark_text_by_word_ids(input_text, inputs["input_ids"][0], start_effect_id, end_effect_id, "#90EE90")  # Light green for effect
+        signal_text = mark_text_by_word_ids(input_text, inputs["input_ids"][0], start_signal_id, end_signal_id, "#FF6347")  # Tomato red for signal
+
+        st.markdown(f"**Cause:**<br>{cause_text}", unsafe_allow_html=True)
+        st.markdown(f"**Effect:**<br>{effect_text}", unsafe_allow_html=True)
+        st.markdown(f"**Signal:**<br>{signal_text}", unsafe_allow_html=True)
+    else:
+        st.warning("Please enter some text before extracting.")
+
+
+
 
 if st.button("Extract1"):
     if input_text:
@@ -207,5 +272,6 @@ if st.button("Extract1"):
             st.markdown(f"<strong>Relation 2:</strong>", unsafe_allow_html=True)
             st.markdown(f"**Cause:** {cause_text2}", unsafe_allow_html=True)
             st.markdown(f"**Effect:** {effect_text2}", unsafe_allow_html=True)
+            st.markdown(f"**Signal:** {signal_text}", unsafe_allow_html=True)
     else:
         st.warning("Please enter some text before extracting.")