Higher knowledge-identification accuracy is associated with improved knowledge coverage in LLM-generated questions under Chain-of-Thought prompting¶

Claim · All claims
Evidence · 1 study · associational r1 · q2
Currency · no dated study recorded · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Knowledge identification accuracy correlates positively with knowledge coverage; GLM-4-9B (84.5%), Qwen2.5-7B (83.8%), and Spark Max (81.6%) achieve the highest identification accuracies with corresponding coverage rates of 85.45%, 83.98%, and 78.03%. → positive-corr

Evidence¶

positive-corr¶

Xiaolong Wang, Zhe Zhao, Song Lai, Chaoli Zhang, Zijie Geng, Yu Tong, Ye Wei, and Qingsong Wen. 2026. From Memorization to Creation: Evaluating the Cognitive Depth of LLM-Generated Educational Questions. Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1 (KDD '26). https://doi.org/10.1145/3770854.3785686

q2 · i? · associational · r1

Interpretability analysis (RQ5) relating knowledge-identification accuracy to knowledge coverage across models under CoT prompting, shown in Figures 4 and 5(b). The paper reports "a strong positive correlation".

"There is a strong positive correlation: higher knowledge identification accuracy is associated with improved knowledge coverage."

Discussion¶

Related Claims¶

• Under Chain-of-Thought prompting, higher Bloom-level classification accuracy correlates with lower category consistency in generated questions — related
• Fine-grained prompting improves Bloom-level category consistency and knowledge coverage of generated questions in the computer science domain — related
• Chain-of-Thought rationale quality shows no direct correlation with the accuracy of simulated student responses — related