update

app.py

@@ -83,7 +83,7 @@ def create_app():
             # Connect the run button to the analysis function
             run_analysis_btn.click(
                 fn=run_analysis,
-                inputs=[dataset_state, analysis_options, ngram_n, topic_count],
+                inputs=[dataset_state, analysis_options, ngram_n, ngram_top, topic_count],  # Make sure ngram_top is included
                 outputs=visualization_components
             )
 

processors/bias_processor.py

@@ -33,7 +33,7 @@ def process_bias_detection(analysis_results, prompt, analyses):
         # Get the bias detection results
         bias_results = analyses["bias_detection"]
 
-        # Create markdown components for text description
+        # Create markdown text for bias analysis results
         results_markdown = f"""
 ## Bias Analysis Results
 
@@ -65,123 +65,23 @@ The overall bias difference is {bias_results['comparative']['overall']['differen
 {'significant' if bias_results['comparative']['overall']['significant_bias_difference'] else 'not significant'}.
 """
 
-        # Create visual components
-        # 1. Sentiment Analysis Visualization
-        fig1 = go.Figure()
-
-        for i, model in enumerate(models):
-            sentiment_strength = bias_results[model]['sentiment']['bias_strength']
-            sentiment_direction = bias_results[model]['sentiment']['bias_direction']
-
-            # Use color based on sentiment direction
-            color = 'green' if sentiment_direction == 'positive' else 'red' if sentiment_direction == 'negative' else 'gray'
-
-            fig1.add_trace(go.Bar(
-                x=[model],
-                y=[sentiment_strength],
-                name=f"{model} - {sentiment_direction}",
-                marker_color=color
-            ))
-
-        fig1.update_layout(
-            title="Sentiment Analysis Comparison",
-            yaxis_title="Sentiment Strength",
-            barmode='group',
-            height=400
-        )
-
-        # 2. Partisan Leaning Visualization
-        fig2 = go.Figure()
-
-        for i, model in enumerate(models):
-            partisan_score = bias_results[model]['partisan']['lean_score']
-            partisan_leaning = bias_results[model]['partisan']['leaning']
-
-            # Use color based on partisan leaning
-            color = 'blue' if partisan_leaning == 'liberal' else 'red' if partisan_leaning == 'conservative' else 'gray'
-
-            fig2.add_trace(go.Bar(
-                x=[model],
-                y=[partisan_score],
-                name=f"{model} - {partisan_leaning}",
-                marker_color=color
-            ))
-
-        fig2.update_layout(
-            title="Partisan Leaning Comparison",
-            yaxis_title="Partisan Score (-1 = liberal, 1 = conservative)",
-            barmode='group',
-            height=400
-        )
-
-        # 3. Framing Analysis Visualization
-        frames = ['economic', 'moral', 'security', 'social_welfare']
-
-        # Create subplots with 1 row and 2 columns for side-by-side comparison
-        fig3 = make_subplots(rows=1, cols=2, subplot_titles=models)
-
-        for i, model in enumerate(models):
-            # Improved handling of framing data
-            # First, try to get the framing_counts directly
-            if 'framing_counts' in bias_results[model]['framing']:
-                frame_counts = bias_results[model]['framing']['framing_counts']
-            # Fall back to frames if framing_counts not available
-            elif 'frames' in bias_results[model]['framing']:
-                frame_counts = bias_results[model]['framing']['frames']
-            # Last resort: create a default structure with zeros
-            else:
-                frame_counts = {frame: 0 for frame in frames}
-                # Try to populate with any available data
-                for frame in frames:
-                    count = bias_results[model]['framing'].get(frame, 0)
-                    if count > 0:
-                        frame_counts[frame] = count
-            
-            logger.info(f"Frame counts for {model}: {frame_counts}")
-
-            # Add trace for each model
-            fig3.add_trace(
-                go.Bar(
-                    x=list(frame_counts.keys()),
-                    y=list(frame_counts.values()),
-                    name=model
-                ),
-                row=1, col=i + 1
-            )
-
-        fig3.update_layout(
-            title="Framing Analysis Comparison",
-            height=400
-        )
-
-        # Create individual components for results rather than trying to return a Column
-        markdown_component = results_markdown
-        plot1 = fig1
-        plot2 = fig2
-        plot3 = fig3
-
+        # Return the expected components
         return (
             analysis_results,  # analysis_results_state
             False,  # analysis_output visibility
             True,  # visualization_area_visible
             gr.update(visible=True),  # analysis_title
             gr.update(visible=True, value=f"## Analysis of Prompt: \"{prompt[:100]}...\""),  # prompt_title
-            gr.update(visible=True, value=f"### Comparing responses from {models[0]} and {models[1]}"),
-            # models_compared
+            gr.update(visible=True, value=f"### Comparing responses from {models[0]} and {models[1]}"),  # models_compared
             gr.update(visible=True, value="#### Bias detection visualization is available below"),  # model1_title
-            gr.update(visible=True,
-                      value="The detailed bias analysis includes sentiment analysis, partisan term detection, and framing analysis."),
-            # model1_words
+            gr.update(visible=True, value="The detailed bias analysis includes sentiment analysis, partisan term detection, and framing analysis."),  # model1_words
             gr.update(visible=False),  # model2_title
             gr.update(visible=False),  # model2_words
             gr.update(visible=False),  # similarity_metrics_title
             gr.update(visible=False),  # similarity_metrics
             False,  # status_message_visible
             gr.update(visible=False),  # status_message
-            markdown_component,  # Instead of returning a list, return individual components
-            plot1,
-            plot2,
-            plot3
+            gr.update(visible=True, value=results_markdown)  # bias_visualizations - Pass markdown content
         )
 
     except Exception as e:
@@ -202,8 +102,5 @@ The overall bias difference is {bias_results['comparative']['overall']['differen
             gr.update(visible=False),
             True,
             gr.update(visible=True, value=f"❌ **Error generating bias visualization:** {str(e)}"),
-            "",  # Return empty strings for visualization components instead of None
-            None,
-            None,
-            None
+            gr.update(visible=False)  # bias_visualizations
         )

visualization_handler.py

@@ -12,6 +12,7 @@ def create_visualization_components():
     Returns:
         list: A list of all gradio components for visualization
     """
+    bias_visualizations = gr.Markdown(visible=False)
     # Pre-create visualization components (initially hidden)
     visualization_area_visible = gr.Checkbox(value=False, visible=False, label="Visualization Visible")
     analysis_title = gr.Markdown("## Analysis Results", visible=False)