Chapter 12: ConceptMap-Text

Chapter 12: ConceptMap-Text — Exploration and Saving

12.1 Step 7: Exploration

A screen for analyzing the built model from multiple perspectives. Multiple tabs provide different viewpoints.

3D Network Graph Tab

The primary visualization of the concept structure model.

Display Elements:

• Nodes (spheres): Concept representative points placed by GNG. Sphere size proportional to assigned data volume
• Edges (lines): MST (Minimum Spanning Tree) connections. Nearby concepts are linked
• Colors: Color-coded by cluster (color mode can be changed)

Operations:

• Rotate: Left mouse button drag
• Zoom: Mouse wheel (scroll)
• Pan (move): Right mouse button drag
• Select node: Click a node to show its detail panel

Display Controls:

Color Mode Details:

• Cluster: Different color for each cluster
• Frequency: Color intensity by data assignment volume. Nodes with more data are darker
• QE (Quantization Error): Model reproduction accuracy. Nodes with high values have lower representativeness
• Each Dimension Value: Color gradient by selected dimension value

Color Scale:

• When color mode is Frequency / QE / dimension value, a gradient bar is displayed
• Drag the handles on the bar to filter the node value range
• Out-of-range nodes are displayed semi-transparently

Snapshot Feature:

• “Capture Camera View” button: Records current viewpoint (rotation angle, zoom, position)
• Recorded viewpoints are listed and can be restored by clicking
• Useful for presentations and reports

Similarity Sort:

• With a node selected, click “Sort by Similarity”
• Nodes closest to the selected node are listed in order of distance
• Useful for discovering related concepts

Cluster Tab

Statistical analysis screen for clusters.

Display:

• Cluster list (name, color, node count, data count)
• Click a cluster to expand details

Cluster Details:

• Feature Profile (bar chart): Shows how each dimension value differs from the overall average. Positive values are above average, negative below
• Node List: List of nodes in the cluster
• Node Selection Filter: Selecting a specific node shows only that node’s assigned data

Reading Feature Profiles:

Example: If the “Technological Innovation” cluster’s feature profile shows +0.8 on the “Abstract ↔ Concrete” dimension, it means nodes in this cluster are 0.8 points more toward “Concrete” than the overall average.

Data Table Tab

View all data in table format.

Features:

• Search: Filter by keyword text input
• Sort: Click column headers to sort
• Pagination: Large datasets split across multiple pages
• Node linkage: Selecting a node in the 3D graph highlights corresponding data rows

AI Chat Tab

Ask AI questions based on the built model’s data.

Operations:

1. Enter a question about the model in the text input field
2. Send to have AI search the model data and respond

Tools Used by AI:

• search_nodes(query): Search nodes by keyword
• get_node_detail(node_id): Get node details
• get_cluster_details(cluster_id): Get cluster overview
• get_table_data(filter): Search original data table

Question Examples:

• “What are the main themes of this model?”
• “Tell me more about Cluster 3’s characteristics”
• “Which nodes are related to ‘user experience’?”
• “Analyze the differences between Cluster 1 and Cluster 4”

12.2 Saving and Loading

Manage model state from the “Save/Load” button at the bottom of the screen.

Save Tab

Save Type Differences:

• Checkpoint: Save in-progress work. Resume later
• Complete: Save completed analysis. Store as final result
• Shared: Save in a shareable format for other users

Operation: Enter each field and click “Save”

Load Tab

• List of previous saves (save date, project name, save name, type)
• “Load” button to restore a save
• “Delete” button to remove a save
• “Load from File” button to restore from an exported file

Rehydration: When loading a saved model, engine session reconstruction (rehydration) is performed automatically. This may take several seconds to tens of seconds.

Export Tab

Operation: Select export format and click “Export”

12.3 Troubleshooting