voidly
Active-learning candidate #1524

South Korea KR · Invalid Date

The classifier predicted 51.0% probability of censorship for South Korea on Invalid Date — sitting just 1.0pp from the 0.5 decision threshold. Your label here closes the loop.

Probability
51%
above 0.5
Distance from 0.5
1.0pp
uncertainty gap
Resolution
resolved
resolved wrong
Consensus need
3+ votes
independent reviewers
Why this candidate was flagged
  • · very close to threshold
  • · already resolved as wrong

Top SHAP contributions

FeatureSHAPDirection
-0.013
+0.009
-0.008

Source: /v1/forecast/KR/7day

Recent incidents in South Korea

Suggest a label

Pick the outcome that actually happened on Invalid Date in South Korea. Anonymous; no login required. Three independent reviewers must agree before a label is promoted into the next retrain dataset.

What actually happened on 2026-05-27#dbscan_flip in KR?

0/2000 characters. The full text is preserved in the audit log; only PII (email, IP) is stripped before consensus.

Stored ONLY in the append-only audit log. Stripped from the consensus file. Never exposed via any read endpoint.

POST /v1/sentinel/active-learning-submit-label

Related