Unverified50% confidenceBenchmarkExact time
It was officially tested on 100 manually labeled English reviews from Chicago businesses (not used for tuning), with 89% agreement on complaint type, 98% on sentiment, and 95% on praise; all 10 serious-issue alerts hit, with 1 false positive out of 93.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/18/2026
First Seen
Valid until: 1/16/2027
Sources
Related Entities
Related Claims
UnverifiedKorean Review Classifier was tested on 100 manually labeled Korean reviews (from Coupang, Naver Shopping, and Naver Place, 47 of which were 1–4 stars), achieving an 88% complaint-type match rate and a 96% sentiment match rate.69% similarUnverifiedIn another evaluation of 330 merchants, the paper scored using deterministic rules with no human raters; only 115 merchants (35%) reached the rankable threshold, and the median score of rankable merchants was between 89 and 90.66% similarUnverifiedIn a specific challenge the author observed, the Metrics tab showed an application with an overall score of 94% that still had 36 major correctness issues.65% similarUnverifiedAmong 19 studies that reported which direction the evaluator's errors lean, 12 lean toward passing work that should have failed, 3 lean in the opposite direction, and 4 are mixed.65% similarUnverifiedIn a standalone test of the "view incomplete" prompt on 100 questions, the wrong-answer rate was 16% without the prompt and dropped to 5% with it.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1059632API
curl https://kongchang.com/api/v1/knowledge/claims/1059632MCP
get_claim(id=1059632)