Understanding the Thematic Network

This visualization maps the intellectual relationships between 18 research themes discovered through AI analysis of article content. Each node represents a distinct thematic cluster, and connections (edges) indicate semantic relationships between themes.

The discovery process: Articles were grouped using unsupervised machine learning (KMeans clustering) based on their semantic embeddings. Themes with high cosine similarity (above 0.70) are connected, revealing how different research areas relate to each other.

What the visualization shows:
• Nodes: Thematic clusters, sized by article count
• Edges: Semantic connections, thickness indicates similarity strength
• Colors: Category domains (HE, SJ, AIL, AIT)
• Labels: AI-generated descriptive theme names

Types of connections:
• Within-domain: Themes in the same category
• Cross-domain: Bridges between different categories
• Cross-language: Links across English, French, and Spanish research

How to explore: Filter by category or source to focus on specific research areas. Drag nodes to rearrange, zoom to examine details, and hover for theme information and sample articles.