← JanPulse  ·  Evidence  ·  Ratings sheet

JanPulse signal extraction — human ratings

Rate each response on three axes. Your numbers are the ground truth the module is measured against, so rate what you actually read, not what you would expect a system to say. No model scores appear on this page, deliberately. Progress saves automatically in this browser, and nothing is uploaded anywhere — you copy your ratings out at the end.

Rated 0 of 0
Valence — pleasant to unpleasant.
Arousal — calm to intense.
Hedging — certain to qualified.
Valence and arousal are independent (Russell's Circumplex Model of Affect). A calm, content response is pleasant with low arousal; fury is unpleasant with high arousal. Judge the text only — not your opinion of the scheme or the speaker. Two decimal places is plenty. If a scale is genuinely unclear, leave it at -- rather than guessing.

Export

Rate every item, put your name in the box at the top, then generate the CSV and send it back to the person who shared this link with you. Rates left alone are exported blank and excluded from the analysis.