What Happens When You Paste a Text
InfraNodus represents a text as a network: the concepts (or entities) are the nodes, and their co-occurrences are the connections. Under the hood (full methodology):
- The text is split into a sequence of lemmas (word roots), and stopwords are removed.
- Words are connected when they appear within a 4-word sliding window — adjacent words get the strongest link, words further apart a weaker one. This mimics how short-term memory holds context.
- Nodes are sized by betweenness centrality — how important a word is for connecting different parts of the discourse.
- A community detection algorithm groups the nodes into topical clusters, shown as colors, laid out with Force-Atlas so related ideas sit close together.
The result is not a word cloud: it is a map of relations. Click any node to see the exact statements behind it, and explore the periphery — the ideas that connect your text to other discourses.
What to Analyze First
Anything works, but the best first graphs come from text you know well, so you can judge what the structure reveals:
- Your notes or journal — paste a few pages of free writing.
- An article or essay draft — yours or one you’re studying.
- A PDF or Markdown files — import directly (Obsidian vaults keep their
[[wiki links]]as connections). - A YouTube video URL — InfraNodus transcribes it automatically.
- A topic you’re curious about — use the AI knowledge graph generator to see how an LLM structures the domain before you add your own thinking.
You can also build a graph by hand as you think: add statements one by one, using #hashtags and @mentions to construct a mental map, or [[wiki links]] for an entity-level ontology (more on ontologies).
Reading Your First Graph in 30 Seconds
Three questions, three answers on screen:
- What is this text about? The colored clusters in the Analytics panel are your topics, named automatically.
- What holds it together? The biggest nodes are the concepts everything else routes through.
- What’s missing? The empty space between distant clusters — the structural gaps — is where Step 2 will generate your first insights.
Don’t analyze anything yet. The point of this step is just to have a graph saved in your account — the next steps do the reading for you.