Concept Extraction
Reveal mode
When you write a note, Relinkia can read it and surface the concepts already present in your text. It is not inventing ideas — it is revealing what is already there, in your own words.
The AI surfaces what's already in your writing. It doesn't add ideas — it names the ones you already expressed.
This matters because concepts become nodes in your graph. The more accurately they reflect your actual vocabulary — the words you use, the distinctions you make — the more useful the graph becomes. Concept extraction is a first draft of that vocabulary, not a final decision.
You validate — you own the vocabulary
After extraction, each suggested concept is presented for your review. You see it, judge it, and decide: does this belong in how I think about this topic?
The concept is added to your graph as a node. It will appear in connection suggestions and in your concept map.
The concept is not added. Relinkia notes that this term is not part of your vocabulary for this context — and adjusts future suggestions accordingly.
If the suggested name is close but not right, rename it. The concept is added under your preferred term.
What rejection teaches
Rejecting a concept suggestion is not a correction — it is calibration. When you say “this is not how I talk about this idea”, Relinkia learns the boundaries of your vocabulary. Over time, suggestions become more aligned with how you actually think.
A system that only sees acceptances gets generic fast. The specificity of your graph comes from the rejections — from the distinctions you actively maintain.
Reject confidently.Every “no” you give makes the next suggestion better. Your rejections are some of the most valuable feedback the system receives.