update

visualization/bow_visualizer.py

@@ -171,8 +171,10 @@ def process_and_visualize_analysis(analysis_results):
                         word_list = [f"{item['word']} ({item['count']})" for item in words[:10]]
                         components.append(gr.Markdown(f"**{model}**: {', '.join(word_list)}"))
                 
-                # Create visualizations but place them inside an accordion
-                with gr.Accordion("📊 View Word Frequency Visualizations", open=False) as bow_accordion:
+                # Create a button to show visualizations
+                def show_bow_visualizations():
+                    visualization_components = []
+                    
                     # Generate model-specific word frequency charts
                     if "important_words" in bow_results:
                         for model_name, words in bow_results.get("important_words", {}).items():
@@ -180,9 +182,9 @@ def process_and_visualize_analysis(analysis_results):
                             
                             # Create bar chart for top words
                             fig = px.bar(df.head(15), x='word', y='count', 
-                                        title=f"Top Words Used by {model_name}",
-                                        labels={'word': 'Word', 'count': 'Frequency'},
-                                        height=400)
+                                      title=f"Top Words Used by {model_name}",
+                                      labels={'word': 'Word', 'count': 'Frequency'},
+                                      height=400)
                             
                             # Improve layout
                             fig.update_layout(
@@ -191,7 +193,7 @@ def process_and_visualize_analysis(analysis_results):
                                 xaxis={'categoryorder':'total descending'}
                             )
                             
-                            components.append(gr.Plot(value=fig))
+                            visualization_components.append(gr.Plot(value=fig))
                     
                     # Visualize differential words
                     models = bow_results.get("models", [])
@@ -237,41 +239,91 @@ def process_and_visualize_analysis(analysis_results):
                                 height=500
                             )
                             
-                            components.append(gr.Plot(value=fig))
+                            visualization_components.append(gr.Plot(value=fig))
+                    
+                    return visualization_components
                 
-                components.append(bow_accordion)
-
-            # Check for N-gram analysis
-            if "ngram_analysis" in analyses:
-                print("Processing N-gram visualization")
-                # Use the dedicated n-gram visualization function
-                with gr.Accordion("📊 View N-gram Analysis Visualizations", open=False) as ngram_accordion:
-                    ngram_components = create_ngram_visualization(
-                        {"analyses": {prompt: {"ngram_analysis": analyses["ngram_analysis"]}}})
-                    components.extend(ngram_components)
-                components.append(ngram_accordion)
-            
-            # Check for Topic Modeling analysis
-            if "topic_modeling" in analyses:
-                print("Processing Topic Modeling visualization")
-                # Use the dedicated topic visualization function
-                with gr.Accordion("📊 View Topic Modeling Visualizations", open=False) as topic_accordion:
-                    topic_components = process_and_visualize_topic_analysis(
-                        {"analyses": {prompt: {"topic_modeling": analyses["topic_modeling"]}}})
-                    components.extend(topic_components)
-                components.append(topic_accordion)
+                # Add a button to show visualizations
+                show_button = gr.Button("📊 Show Word Frequency Visualizations")
+                visualization_container = gr.Column(visible=False)
+                
+                def show_visualizations():
+                    return gr.Column(visible=True)
                 
-            # Check for Bias Detection analysis
-            if "bias_detection" in analyses:
-                print("Processing Bias Detection visualization")
-                # Use the dedicated bias visualization function
-                from visualization.bias_visualizer import process_and_visualize_bias_analysis
+                show_button.click(
+                    fn=show_visualizations,
+                    outputs=visualization_container
+                )
                 
-                with gr.Accordion("📊 View Bias Analysis Visualizations", open=False) as bias_accordion:
-                    bias_components = process_and_visualize_bias_analysis(
-                        {"analyses": {prompt: {"bias_detection": analyses["bias_detection"]}}})
-                    components.extend(bias_components)
-                components.append(bias_accordion)
+                with visualization_container:
+                    # Generate model-specific word frequency charts
+                    if "important_words" in bow_results:
+                        for model_name, words in bow_results.get("important_words", {}).items():
+                            df = pd.DataFrame(words)
+                            
+                            # Create bar chart for top words
+                            fig = px.bar(df.head(15), x='word', y='count', 
+                                      title=f"Top Words Used by {model_name}",
+                                      labels={'word': 'Word', 'count': 'Frequency'},
+                                      height=400)
+                            
+                            # Improve layout
+                            fig.update_layout(
+                                xaxis_title="Word",
+                                yaxis_title="Frequency",
+                                xaxis={'categoryorder':'total descending'}
+                            )
+                            
+                            gr.Plot(value=fig)
+                    
+                    # Visualize differential words
+                    models = bow_results.get("models", [])
+                    diff_words = bow_results.get("differential_words", [])
+                    word_matrix = bow_results.get("word_count_matrix", {})
+                    
+                    if len(models) >= 2 and diff_words and word_matrix:
+                        model1, model2 = models[0], models[1]
+                        diff_data = []
+                        
+                        for word in diff_words[:15]:
+                            if word in word_matrix:
+                                counts = word_matrix[word]
+                                diff_data.append({
+                                    "word": word,
+                                    model1: counts.get(model1, 0),
+                                    model2: counts.get(model2, 0)
+                                })
+                        
+                        if diff_data:
+                            diff_df = pd.DataFrame(diff_data)
+                            
+                            # Create grouped bar chart
+                            fig = go.Figure()
+                            fig.add_trace(go.Bar(
+                                x=diff_df['word'],
+                                y=diff_df[model1],
+                                name=model1,
+                                marker_color='indianred'
+                            ))
+                            fig.add_trace(go.Bar(
+                                x=diff_df['word'],
+                                y=diff_df[model2],
+                                name=model2,
+                                marker_color='lightsalmon'
+                            ))
+                            
+                            fig.update_layout(
+                                title="Word Frequency Comparison",
+                                xaxis_title="Word",
+                                yaxis_title="Frequency",
+                                barmode='group',
+                                height=500
+                            )
+                            
+                            gr.Plot(value=fig)
+                
+                components.append(show_button)
+                components.append(visualization_container)
 
         if not components:
             components.append(gr.Markdown("No visualization components could be created from the analysis results."))
@@ -282,5 +334,4 @@ def process_and_visualize_analysis(analysis_results):
         import traceback
         error_msg = f"Visualization error: {str(e)}\n{traceback.format_exc()}"
         print(error_msg)
-        return [gr.Markdown(f"**Error during visualization:**\n\n```\n{error_msg}\n```")]
-
+        return [gr.Markdown(f"**Error during visualization:**\n\n```\n{error_msg}\n```")]