The paper "Understanding and improving fairness in cognitive diagnosis" by Zheng ZHANG et al. addresses the critical issue of fairness in cognitive diagnosis (CD), a key application of artificial intelligence in education. The authors explore whether existing CD models are influenced by sensitive attributes, such as region or gender, and propose methods to mitigate this impact. They empirically demonstrate that several well-known CD methods can lead to unfair performance, varying across different models. Theoretical analysis reveals that model complexity contributes to these unfair outcomes. To address this, the authors introduce FairCD, a fairness-aware CD framework that decomposes student proficiency into bias proficiency and fair proficiency. This framework ensures that fairness is independent of sensitive attributes and is validated through extensive experiments on the PISA dataset. The paper highlights the importance of fairness in educational settings and provides a practical solution to ensure equitable diagnostic results.The paper "Understanding and improving fairness in cognitive diagnosis" by Zheng ZHANG et al. addresses the critical issue of fairness in cognitive diagnosis (CD), a key application of artificial intelligence in education. The authors explore whether existing CD models are influenced by sensitive attributes, such as region or gender, and propose methods to mitigate this impact. They empirically demonstrate that several well-known CD methods can lead to unfair performance, varying across different models. Theoretical analysis reveals that model complexity contributes to these unfair outcomes. To address this, the authors introduce FairCD, a fairness-aware CD framework that decomposes student proficiency into bias proficiency and fair proficiency. This framework ensures that fairness is independent of sensitive attributes and is validated through extensive experiments on the PISA dataset. The paper highlights the importance of fairness in educational settings and provides a practical solution to ensure equitable diagnostic results.