Update app.py

app.py

@@ -13,59 +13,151 @@ def load_model(model_name, model_class, is_bc=False):
     model = model_class.from_pretrained(model_name, num_labels=3 if not is_bc else 2)
     return tokenizer, model
 
-# UI Configuration
+# Set up page configuration and custom CSS for a modern, clean look
 st.set_page_config(page_title="SemViQA Demo", layout="wide")
 
 st.markdown("""
     <style>
-        .big-title { font-size: 36px; font-weight: bold; color: #4A90E2; text-align: center; }
-        .sub-title { font-size: 20px; color: #666; text-align: center; }
-        .stButton>button { background-color: #4CAF50; color: white; font-size: 16px; width: 100%; border-radius: 8px; padding: 10px; }
-        .stTextArea textarea { font-size: 16px; }
-        .result-box { background-color: #f9f9f9; padding: 20px; border-radius: 10px; box-shadow: 0px 4px 8px rgba(0, 0, 0, 0.1); }
+        .big-title { 
+            font-size: 36px; 
+            font-weight: bold; 
+            color: #4A90E2; 
+            text-align: center; 
+            margin-top: 20px;
+        }
+        .sub-title { 
+            font-size: 20px; 
+            color: #666; 
+            text-align: center; 
+            margin-bottom: 20px;
+        }
+        .stButton>button { 
+            background-color: #4CAF50; 
+            color: white; 
+            font-size: 16px; 
+            width: 100%; 
+            border-radius: 8px; 
+            padding: 10px; 
+        }
+        .stTextArea textarea { 
+            font-size: 16px; 
+        }
+        .result-box { 
+            background-color: #f9f9f9; 
+            padding: 20px; 
+            border-radius: 10px; 
+            box-shadow: 0px 4px 8px rgba(0, 0, 0, 0.1); 
+            margin-top: 20px;
+        }
+        .verdict { 
+            font-size: 24px; 
+            font-weight: bold; 
+            margin: 0;
+            display: flex;
+            align-items: center;
+        }
+        .verdict-icon { 
+            margin-right: 10px;
+        }
     </style>
 """, unsafe_allow_html=True)
 
-st.markdown("<p class='big-title'>🔍 SemViQA: A Semantic Question Answering System for Vietnamese Information Fact-Checking</p>", unsafe_allow_html=True)
+st.markdown("<p class='big-title'>SemViQA: Semantic Question Answering System for Vietnamese Fact-Checking</p>", unsafe_allow_html=True)
 st.markdown("<p class='sub-title'>Enter a claim and context to verify its accuracy</p>", unsafe_allow_html=True)
 
-# Sidebar - Configuration Settings
+# Sidebar: Settings and additional features
 with st.sidebar.expander("⚙️ Settings", expanded=False):
-    tfidf_threshold = st.slider("🔧 TF-IDF Threshold", 0.0, 1.0, 0.5, 0.01)
-    length_ratio_threshold = st.slider("📏 Length Ratio Threshold", 0.1, 1.0, 0.5, 0.01)
-    qatc_model_name = st.selectbox("🤖 QATC Model", ["SemViQA/qatc-infoxlm-viwikifc","SemViQA/qatc-infoxlm-isedsc01","SemViQA/qatc-vimrc-viwikifc","SemViQA/qatc-vimrc-isedsc01"])
-    bc_model_name = st.selectbox("🏷️ Binary Classification Model", ["SemViQA/bc-xlmr-viwikifc","SemViQA/bc-xlmr-isedsc01","SemViQA/bc-infoxlm-viwikifc","SemViQA/bc-infoxlm-isedsc01","SemViQA/bc-erniem-viwikifc","SemViQA/bc-erniem-isedsc01"])
-    tc_model_name = st.selectbox("📊 Three-Class Classification Model", ["SemViQA/tc-xlmr-viwikifc","SemViQA/tc-xlmr-isedsc01","SemViQA/tc-infoxlm-viwikifc","SemViQA/tc-infoxlm-isedsc01","SemViQA/tc-erniem-viwikifc","SemViQA/tc-erniem-isedsc01"])
+    tfidf_threshold = st.slider("TF-IDF Threshold", 0.0, 1.0, 0.5, 0.01)
+    length_ratio_threshold = st.slider("Length Ratio Threshold", 0.1, 1.0, 0.5, 0.01)
+    qatc_model_name = st.selectbox("QATC Model", [
+        "SemViQA/qatc-infoxlm-viwikifc",
+        "SemViQA/qatc-infoxlm-isedsc01",
+        "SemViQA/qatc-vimrc-viwikifc",
+        "SemViQA/qatc-vimrc-isedsc01"
+    ])
+    bc_model_name = st.selectbox("Binary Classification Model", [
+        "SemViQA/bc-xlmr-viwikifc",
+        "SemViQA/bc-xlmr-isedsc01",
+        "SemViQA/bc-infoxlm-viwikifc",
+        "SemViQA/bc-infoxlm-isedsc01",
+        "SemViQA/bc-erniem-viwikifc",
+        "SemViQA/bc-erniem-isedsc01"
+    ])
+    tc_model_name = st.selectbox("Three-Class Classification Model", [
+        "SemViQA/tc-xlmr-viwikifc",
+        "SemViQA/tc-xlmr-isedsc01",
+        "SemViQA/tc-infoxlm-viwikifc",
+        "SemViQA/tc-infoxlm-isedsc01",
+        "SemViQA/tc-erniem-viwikifc",
+        "SemViQA/tc-erniem-isedsc01"
+    ])
+    show_details = st.checkbox("Show probability details", value=False)
 
-# Load selected models
+# Initialize verification history in session state
+if 'history' not in st.session_state:
+    st.session_state.history = []
+
+# Load the selected models
 tokenizer_qatc, model_qatc = load_model(qatc_model_name, QATCForQuestionAnswering)
 tokenizer_bc, model_bc = load_model(bc_model_name, ClaimModelForClassification, is_bc=True)
 tokenizer_tc, model_tc = load_model(tc_model_name, ClaimModelForClassification)
 
-# User Input Fields
-claim = st.text_area("✍️ Enter Claim", "Vietnam is a country in Southeast Asia.")
-context = st.text_area("📖 Enter Context", "Vietnam is a country located in Southeast Asia, covering an area of over 331,000 km² with a population of more than 98 million people.")
+# User input fields
+claim = st.text_area("Enter Claim", "Vietnam is a country in Southeast Asia.")
+context = st.text_area("Enter Context", "Vietnam is a country located in Southeast Asia, covering an area of over 331,000 km² with a population of more than 98 million people.")
+
+# Define icon mapping for each verdict label
+verdict_icons = {
+    "SUPPORTED": "✅",
+    "REFUTED": "❌",
+    "NEI": "⚠️"
+}
+
+if st.button("Verify"):
+    with st.spinner("Verifying..."):
+        # Extract evidence
+        evidence = extract_evidence_tfidf_qatc(
+            claim, context, model_qatc, tokenizer_qatc, "cuda" if torch.cuda.is_available() else "cpu",
+            confidence_threshold=tfidf_threshold, length_ratio_threshold=length_ratio_threshold
+        )
+        
+        # Classify the claim
+        verdict = "NEI"
+        prob3class, pred_tc = classify_claim(claim, evidence, model_tc, tokenizer_tc, "cuda" if torch.cuda.is_available() else "cpu")
+        
+        details = ""
+        if pred_tc != 0:
+            prob2class, pred_bc = classify_claim(claim, evidence, model_bc, tokenizer_bc, "cuda" if torch.cuda.is_available() else "cpu")
+            if pred_bc == 0:
+                verdict = "SUPPORTED"
+            elif prob2class > prob3class:
+                verdict = "REFUTED"
+            else:
+                verdict = ["NEI", "SUPPORTED", "REFUTED"][pred_tc]
+            if show_details:
+                details = f"<p><strong>3-Class Probability:</strong> {prob3class:.2f} - <strong>2-Class Probability:</strong> {prob2class:.2f}</p>"
+        
+        # Save the verification record in session history
+        st.session_state.history.append({
+            "claim": claim,
+            "evidence": evidence,
+            "verdict": verdict
+        })
+        
+        # Display the results with icon and label (without extra "Verdict:" text)
+        st.markdown(f"""
+            <div class='result-box'>
+                <h3>Result</h3>
+                <p><strong>Evidence:</strong> {evidence}</p>
+                <p class='verdict'><span class='verdict-icon'>{verdict_icons.get(verdict, '')}</span>{verdict}</p>
+                {details}
+            </div>
+        """, unsafe_allow_html=True)
 
-if st.button("🔎 Verify"):
-    # Extract evidence
-    evidence = extract_evidence_tfidf_qatc(
-        claim, context, model_qatc, tokenizer_qatc, "cuda" if torch.cuda.is_available() else "cpu",
-        confidence_threshold=tfidf_threshold, length_ratio_threshold=length_ratio_threshold
-    )
-    
-    # Claim Classification
-    verdict = "NEI"
-    prob3class, pred_tc = classify_claim(claim, evidence, model_tc, tokenizer_tc, "cuda" if torch.cuda.is_available() else "cpu")
-    
-    if pred_tc != 0:
-        prob2class, pred_bc = classify_claim(claim, evidence, model_bc, tokenizer_bc, "cuda" if torch.cuda.is_available() else "cpu")
-        verdict = "SUPPORTED" if pred_bc == 0 else "REFUTED" if prob2class > prob3class else ["NEI", "SUPPORTED", "REFUTED"][pred_tc]
-    
-    # Display Results
-    st.markdown(f"""
-        <div class='result-box'>
-            <h3>📌 Result</h3>
-            <p><strong>🔍 Evidence:</strong> {evidence}</p>
-            <p><strong>✅ Verdict:</strong> {verdict}</p>
-        </div>
-    """, unsafe_allow_html=True)
+# Display verification history in the sidebar
+with st.sidebar.expander("Verification History", expanded=False):
+    if st.session_state.history:
+        for idx, record in enumerate(reversed(st.session_state.history), 1):
+            st.markdown(f"**{idx}. Claim:** {record['claim']}  \n**Result:** {verdict_icons.get(record['verdict'], '')} {record['verdict']}")
+    else:
+        st.write("No verification history yet.")