Community-reaction rankings Built
A derived analytic product over the corpus: the reply-driven community-reaction classifier ranks every post by the emotional vocabulary its replies drew - the room grading the post. Seven axes, ranked to JSONL on disk. This is meta about the corpus, not a raw possession.
Funny - posts the room laughed at Held on disk
Posts certified funny by their replies ('that's funny', 'LOL', 'made me laugh'), echo-guarded against replies that merely re-paste the parent.
Work-area path:
reactions/funny-candidates.jsonlSad - the elegiac posts Held on disk
Posts the community met with grief or mourning in reply - the memorial and loss threads.
Work-area path:
reactions/sad-candidates.jsonlMean - posts that drew a cutting reply Held on disk
Posts whose replies turned cutting or contemptuous; cross-filterable against the flame/politics content tags on the parent.
Work-area path:
reactions/mean-candidates.jsonlAngry - posts that sparked heat Held on disk
Posts whose replies expressed anger or outrage - the arguments and provocations.
Work-area path:
reactions/angry-candidates.jsonlGrateful - posts the room thanked Held on disk
Posts met with thanks or appreciation in reply - the helpful answers, the shared tapes, the generous acts.
Work-area path:
reactions/grateful-candidates.jsonlWow - posts that awed the room Held on disk
Posts whose replies expressed astonishment or admiration - the standout stories and reveals.
Work-area path:
reactions/wow-candidates.jsonlAgree - posts the room affirmed Held on disk
Posts met with agreement or affirmation in reply - the well-put opinions and shared sentiments.
Work-area path:
reactions/agree-candidates.jsonl
Draws on: The rec.music.gdead Usenet corpus