Unverified50% confidenceFactExact time
When multiple annotators give different judgments, engineering practice typically compresses the results into a single label through aggregation methods such as voting or averaging, or treats the disagreement as noise to be discarded.
1
Sources
50%
Confidence
Long-term
Relevance
10/9/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLLM evaluation is more objective than human evaluation, as it can score candidates consistently within a unified evaluation framework, avoiding the unfairness caused by cognitive biases such as fatigue, mood, and the primacy effect in human interviewers.68% similarUnverified多评判者投票系统在多个存在不同偏差的评判者独立评分时偏差方向相互抵消，聚合结论可靠性高于单一评判者67% similarUnverifiedTreating the judgments of a particular group as correct by default while discarding other reasonable judgments as noise may be a mechanism of implicit bias amplification.67% similarUnverified'Measurement Bias' in questionnaire design refers to the way question wording, option order, and scale extremity systematically influence responses, skewing the distribution toward the direction preset by the designer.66% similarUnverified单一LLM评判者存在位置偏见（倾向于给出现在前面的答案更高评分）和自我增强偏见（倾向于给自家风格的输出打高分）66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/990460API
curl https://kongchang.com/api/v1/knowledge/claims/990460MCP
get_claim(id=990460)