A want enters as an intent, its meaning is recovered, it becomes a typed claim, and a human labels it. Each stage hands off to the next.
{{ f.label }}
{{ f.n }}
{{ f.of }}
{{ f.note }}
Waiting on a human
{{ w.n }}
{{ w.title }}
{{ w.note }}
→
Labels so far
{{ lab.correct }}
correct
{{ lab.incorrect }}
incorrect
{{ lab.rest }}
unlabelled
No corrections recorded yet. When you mark a claim incorrect, the axis you pick shows up here — that is the training signal.
Claims by type
Relations found between claims
Dashed in the graph means the agent proposed it and nobody has confirmed it yet.