feat: enhance verification interface with message review and feedback functionality

- Introduce a new message review section for displaying patient messages and classification results
- Add feedback buttons for users to confirm or correct classifications
- Implement correction section for users to submit corrections with optional notes
- Update session statistics display to reflect progress and accuracy
- Bind new event handlers for feedback and correction actions to improve user interaction

src/interface/enhanced_verification_interface.py

@@ -238,7 +238,7 @@ class EnhancedVerificationInterface:
                                     label="Dataset Details"
                                 )
                         
-                        # Verification interface (initially hidden)
+                        # Verification setup section (initially hidden)
                         verification_section = gr.Row(visible=False)
                         with verification_section:
                             with gr.Column():
@@ -266,12 +266,74 @@ class EnhancedVerificationInterface:
                                     label="Progress"
                                 )
                         
+                        # Message review section (initially hidden)
+                        message_review_section = gr.Row(visible=False)
+                        with message_review_section:
+                            with gr.Column(scale=2):
+                                gr.Markdown("## 📝 Message Review")
+                                
+                                # Current message display
+                                current_message_display = gr.Textbox(
+                                    label="Patient Message",
+                                    interactive=False,
+                                    lines=4
+                                )
+                                
+                                # Classification result
+                                with gr.Row():
+                                    classifier_decision_display = gr.Markdown(
+                                        "🔄 Loading...",
+                                        label="Classifier Decision"
+                                    )
+                                    classifier_confidence_display = gr.Markdown(
+                                        "",
+                                        label="Confidence"
+                                    )
+                                
+                                # Feedback buttons
+                                gr.Markdown("### Is this classification correct?")
+                                with gr.Row():
+                                    correct_btn = gr.Button("✓ Correct", variant="primary", scale=1)
+                                    incorrect_btn = gr.Button("✗ Incorrect", variant="stop", scale=1)
+                                
+                                # Correction section (initially hidden)
+                                correction_section = gr.Row(visible=False)
+                                with correction_section:
+                                    with gr.Column():
+                                        gr.Markdown("### Select correct classification:")
+                                        correction_selector = gr.Radio(
+                                            choices=["green", "yellow", "red"],
+                                            label="Correct Classification",
+                                            value=None
+                                        )
+                                        correction_notes = gr.Textbox(
+                                            label="Notes (Optional)",
+                                            placeholder="Why is this incorrect?",
+                                            lines=2
+                                        )
+                                        with gr.Row():
+                                            submit_correction_btn = gr.Button("✓ Submit Correction", variant="primary")
+                                            cancel_correction_btn = gr.Button("✗ Cancel", variant="secondary")
+                            
+                            with gr.Column(scale=1):
+                                gr.Markdown("## 📊 Session Statistics")
+                                session_stats_display = gr.Markdown(
+                                    "**Progress:** 0/0\n**Correct:** 0\n**Incorrect:** 0\n**Accuracy:** 0%"
+                                )
+                                
+                                # Navigation
+                                with gr.Row():
+                                    skip_btn = gr.Button("⏭️ Skip", variant="secondary")
+                                    finish_btn = gr.Button("🏁 Finish Session", variant="primary")
+                        
                         # Status message for dataset interface
                         dataset_status_message = gr.Markdown("", visible=True)
                         
                         # Dataset state
                         current_dataset_state = gr.State(value=None)
                         verification_session_state = gr.State(value=None)
+                        current_message_index_state = gr.State(value=0)
+                        current_classification_state = gr.State(value=None)
                 
                 manual_input_interface = gr.Row(visible=False)
                 with manual_input_interface:
@@ -755,29 +817,256 @@ class EnhancedVerificationInterface:
                     if not current_dataset:
                         return (
                             None,  # verification_session_state
+                            gr.Row(visible=False),  # message_review_section
+                            "",  # current_message_display
+                            "",  # classifier_decision_display
+                            "",  # classifier_confidence_display
+                            "",  # session_stats_display
+                            0,   # current_message_index_state
+                            None,  # current_classification_state
                             "❌ No dataset selected"  # dataset_status_message
                         )
                     
+                    if not verifier_name or not verifier_name.strip():
+                        return (
+                            None,  # verification_session_state
+                            gr.Row(visible=False),  # message_review_section
+                            "",  # current_message_display
+                            "",  # classifier_decision_display
+                            "",  # classifier_confidence_display
+                            "",  # session_stats_display
+                            0,   # current_message_index_state
+                            None,  # current_classification_state
+                            "❌ Please enter your name"  # dataset_status_message
+                        )
+                    
                     success, message, session = dataset_controller.start_verification_session(
                         current_dataset, verifier_name
                     )
                     
                     if success:
-                        return (
-                            session,  # verification_session_state
-                            message   # dataset_status_message
-                        )
+                        # Get first message for verification
+                        first_message, classification_result = dataset_controller.get_current_message_for_verification()
+                        
+                        if first_message:
+                            # Format classification display
+                            decision = classification_result.get('decision', 'unknown').upper()
+                            confidence = classification_result.get('confidence', 0) * 100
+                            
+                            decision_colors = {'GREEN': '🟢', 'YELLOW': '🟡', 'RED': '🔴'}
+                            decision_badge = f"{decision_colors.get(decision, '❓')} **{decision}**"
+                            confidence_text = f"Confidence: **{confidence:.1f}%**"
+                            
+                            stats_text = f"""**Progress:** 1/{len(current_dataset.messages)}
+**Correct:** 0
+**Incorrect:** 0
+**Accuracy:** N/A"""
+                            
+                            return (
+                                session,  # verification_session_state
+                                gr.Row(visible=True),  # message_review_section
+                                first_message.text,  # current_message_display
+                                decision_badge,  # classifier_decision_display
+                                confidence_text,  # classifier_confidence_display
+                                stats_text,  # session_stats_display
+                                0,   # current_message_index_state
+                                classification_result,  # current_classification_state
+                                message  # dataset_status_message
+                            )
+                        else:
+                            return (
+                                session,  # verification_session_state
+                                gr.Row(visible=False),  # message_review_section
+                                "",  # current_message_display
+                                "",  # classifier_decision_display
+                                "",  # classifier_confidence_display
+                                "",  # session_stats_display
+                                0,   # current_message_index_state
+                                None,  # current_classification_state
+                                "❌ No messages in dataset"  # dataset_status_message
+                            )
                     else:
                         return (
-                            None,     # verification_session_state
-                            message   # dataset_status_message
+                            None,  # verification_session_state
+                            gr.Row(visible=False),  # message_review_section
+                            "",  # current_message_display
+                            "",  # classifier_decision_display
+                            "",  # classifier_confidence_display
+                            "",  # session_stats_display
+                            0,   # current_message_index_state
+                            None,  # current_classification_state
+                            message  # dataset_status_message
                         )
                 except Exception as e:
                     return (
                         None,  # verification_session_state
+                        gr.Row(visible=False),  # message_review_section
+                        "",  # current_message_display
+                        "",  # classifier_decision_display
+                        "",  # classifier_confidence_display
+                        "",  # session_stats_display
+                        0,   # current_message_index_state
+                        None,  # current_classification_state
                         f"❌ Error starting verification: {str(e)}"  # dataset_status_message
                     )
             
+            def on_correct_feedback(current_dataset, session, msg_index, classification):
+                """Handle correct classification feedback."""
+                try:
+                    success, message, stats = dataset_controller.submit_verification_feedback(True)
+                    
+                    if success and not stats.get('is_complete', False):
+                        # Get next message
+                        next_message, next_classification = dataset_controller.get_current_message_for_verification()
+                        
+                        if next_message:
+                            decision = next_classification.get('decision', 'unknown').upper()
+                            confidence = next_classification.get('confidence', 0) * 100
+                            
+                            decision_colors = {'GREEN': '🟢', 'YELLOW': '🟡', 'RED': '🔴'}
+                            decision_badge = f"{decision_colors.get(decision, '❓')} **{decision}**"
+                            confidence_text = f"Confidence: **{confidence:.1f}%**"
+                            
+                            stats_text = f"""**Progress:** {stats['processed'] + 1}/{stats['total']}
+**Correct:** {stats['correct']}
+**Incorrect:** {stats['incorrect']}
+**Accuracy:** {stats['accuracy']:.1f}%"""
+                            
+                            return (
+                                next_message.text,  # current_message_display
+                                decision_badge,  # classifier_decision_display
+                                confidence_text,  # classifier_confidence_display
+                                stats_text,  # session_stats_display
+                                msg_index + 1,  # current_message_index_state
+                                next_classification,  # current_classification_state
+                                gr.Row(visible=False),  # correction_section
+                                message  # dataset_status_message
+                            )
+                    
+                    # Session complete
+                    stats_text = f"""**🎉 Session Complete!**
+**Total:** {stats['processed']}
+**Correct:** {stats['correct']}
+**Incorrect:** {stats['incorrect']}
+**Final Accuracy:** {stats['accuracy']:.1f}%"""
+                    
+                    return (
+                        "✅ Verification session complete!",  # current_message_display
+                        "🎉 **COMPLETE**",  # classifier_decision_display
+                        "",  # classifier_confidence_display
+                        stats_text,  # session_stats_display
+                        msg_index,  # current_message_index_state
+                        None,  # current_classification_state
+                        gr.Row(visible=False),  # correction_section
+                        f"✅ Session complete! Accuracy: {stats['accuracy']:.1f}%"  # dataset_status_message
+                    )
+                except Exception as e:
+                    return (
+                        gr.update(),  # current_message_display
+                        gr.update(),  # classifier_decision_display
+                        gr.update(),  # classifier_confidence_display
+                        gr.update(),  # session_stats_display
+                        msg_index,  # current_message_index_state
+                        classification,  # current_classification_state
+                        gr.Row(visible=False),  # correction_section
+                        f"❌ Error: {str(e)}"  # dataset_status_message
+                    )
+            
+            def on_incorrect_feedback():
+                """Show correction section."""
+                return (
+                    gr.Row(visible=True),  # correction_section
+                    "⚠️ Please select the correct classification"  # dataset_status_message
+                )
+            
+            def on_cancel_correction():
+                """Cancel correction."""
+                return (
+                    gr.Row(visible=False),  # correction_section
+                    "❌ Correction cancelled"  # dataset_status_message
+                )
+            
+            def on_submit_correction(current_dataset, session, msg_index, classification, correction, notes):
+                """Submit correction feedback."""
+                try:
+                    if not correction:
+                        return (
+                            gr.update(),  # current_message_display
+                            gr.update(),  # classifier_decision_display
+                            gr.update(),  # classifier_confidence_display
+                            gr.update(),  # session_stats_display
+                            msg_index,  # current_message_index_state
+                            classification,  # current_classification_state
+                            gr.Row(visible=True),  # correction_section
+                            "❌ Please select a classification"  # dataset_status_message
+                        )
+                    
+                    success, message, stats = dataset_controller.submit_verification_feedback(False, correction, notes)
+                    
+                    if success and not stats.get('is_complete', False):
+                        # Get next message
+                        next_message, next_classification = dataset_controller.get_current_message_for_verification()
+                        
+                        if next_message:
+                            decision = next_classification.get('decision', 'unknown').upper()
+                            confidence = next_classification.get('confidence', 0) * 100
+                            
+                            decision_colors = {'GREEN': '🟢', 'YELLOW': '🟡', 'RED': '🔴'}
+                            decision_badge = f"{decision_colors.get(decision, '❓')} **{decision}**"
+                            confidence_text = f"Confidence: **{confidence:.1f}%**"
+                            
+                            stats_text = f"""**Progress:** {stats['processed'] + 1}/{stats['total']}
+**Correct:** {stats['correct']}
+**Incorrect:** {stats['incorrect']}
+**Accuracy:** {stats['accuracy']:.1f}%"""
+                            
+                            return (
+                                next_message.text,  # current_message_display
+                                decision_badge,  # classifier_decision_display
+                                confidence_text,  # classifier_confidence_display
+                                stats_text,  # session_stats_display
+                                msg_index + 1,  # current_message_index_state
+                                next_classification,  # current_classification_state
+                                gr.Row(visible=False),  # correction_section
+                                message  # dataset_status_message
+                            )
+                    
+                    # Session complete
+                    stats_text = f"""**🎉 Session Complete!**
+**Total:** {stats['processed']}
+**Correct:** {stats['correct']}
+**Incorrect:** {stats['incorrect']}
+**Final Accuracy:** {stats['accuracy']:.1f}%"""
+                    
+                    return (
+                        "✅ Verification session complete!",  # current_message_display
+                        "🎉 **COMPLETE**",  # classifier_decision_display
+                        "",  # classifier_confidence_display
+                        stats_text,  # session_stats_display
+                        msg_index,  # current_message_index_state
+                        None,  # current_classification_state
+                        gr.Row(visible=False),  # correction_section
+                        f"✅ Session complete! Accuracy: {stats['accuracy']:.1f}%"  # dataset_status_message
+                    )
+                except Exception as e:
+                    return (
+                        gr.update(),  # current_message_display
+                        gr.update(),  # classifier_decision_display
+                        gr.update(),  # classifier_confidence_display
+                        gr.update(),  # session_stats_display
+                        msg_index,  # current_message_index_state
+                        classification,  # current_classification_state
+                        gr.Row(visible=True),  # correction_section
+                        f"❌ Error: {str(e)}"  # dataset_status_message
+                    )
+            
+            def on_finish_session(session):
+                """Finish verification session."""
+                return (
+                    gr.Row(visible=False),  # message_review_section
+                    "✅ Session finished. You can start a new verification or select another dataset."  # dataset_status_message
+                )
+            
             # Bind dataset interface events
             dataset_selector.change(
                 on_dataset_selection_change,
@@ -794,7 +1083,70 @@ class EnhancedVerificationInterface:
             start_verification_btn.click(
                 on_start_verification,
                 inputs=[current_dataset_state, verifier_name_input],
-                outputs=[verification_session_state, dataset_status_message]
+                outputs=[
+                    verification_session_state,
+                    message_review_section,
+                    current_message_display,
+                    classifier_decision_display,
+                    classifier_confidence_display,
+                    session_stats_display,
+                    current_message_index_state,
+                    current_classification_state,
+                    dataset_status_message
+                ]
+            )
+            
+            correct_btn.click(
+                on_correct_feedback,
+                inputs=[current_dataset_state, verification_session_state, current_message_index_state, current_classification_state],
+                outputs=[
+                    current_message_display,
+                    classifier_decision_display,
+                    classifier_confidence_display,
+                    session_stats_display,
+                    current_message_index_state,
+                    current_classification_state,
+                    correction_section,
+                    dataset_status_message
+                ]
+            )
+            
+            incorrect_btn.click(
+                on_incorrect_feedback,
+                outputs=[correction_section, dataset_status_message]
+            )
+            
+            cancel_correction_btn.click(
+                on_cancel_correction,
+                outputs=[correction_section, dataset_status_message]
+            )
+            
+            submit_correction_btn.click(
+                on_submit_correction,
+                inputs=[
+                    current_dataset_state,
+                    verification_session_state,
+                    current_message_index_state,
+                    current_classification_state,
+                    correction_selector,
+                    correction_notes
+                ],
+                outputs=[
+                    current_message_display,
+                    classifier_decision_display,
+                    classifier_confidence_display,
+                    session_stats_display,
+                    current_message_index_state,
+                    current_classification_state,
+                    correction_section,
+                    dataset_status_message
+                ]
+            )
+            
+            finish_btn.click(
+                on_finish_session,
+                inputs=[verification_session_state],
+                outputs=[message_review_section, dataset_status_message]
             )
         
         return interface