Platform for collecting, tracking, and rating crowd-sourced model interviews and conversations.
Research-critical data with flexible schemas:
data/submissions/{submissionId}/- Submission metadata, messages, ratingsdata/topics.jsonl- Research topicsdata/criteria.jsonl- Evaluation criteriadata/users/{userId}.jsonl- User records
Social/query-heavy with stable schemas:
selections- Text ranges/annotations in submissionscomments- Threaded comments on submissions/selectionsratings- Linked to comments and criteria
- Calls ARC API for certified conversation data
- Independent user registration (not shared with ARC)
npm install
npm run devSubmissions contain a tree of messages (branching chat conversations). Researchers create selections (text ranges), comment on them, and rate them against criteria.
See src/types/ for detailed schemas.