Unverified50% confidenceFactExact time
ComFair is a configurable fairness benchmark that converts existing fairness data into 479,066 controlled two-attribute and multi-attribute instances and pairs each augmented instance with a task-equivalent source instance.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/27/2026
First Seen
Valid until: 1/25/2027
Sources
Related Entities
Related Claims
UnverifiedThe paper "Beyond Single-Attribute Auditing: A Configurable Multi-Attribute Benchmark for Intersectional Fairness in LLMs" points out that existing fairness benchmarks mostly use single-attribute auditing and cannot capture model behavior bias under joint multi-attribute distributions.64% similarUnverifiedAccountability requires data controllers to be able to demonstrate that their processing activities comply, not merely to claim compliance.63% similarUnverifiedThe counterexample is constructed using independent fair labels (each sampled label is equally likely to be positive or negative) and a permutation-equivariant learner.63% similarUnverified该数据集涵盖正确性bug、安全问题、性能权衡和真实的风格判断等多种编程质量维度61% similarUnverified等价类划分将输入数据按有效/无效归类，从每个等价类中各取一个代表性数据进行测试，以在有限用例数量下最大化覆盖率60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1101983API
curl https://kongchang.com/api/v1/knowledge/claims/1101983MCP
get_claim(id=1101983)