cs.CL提交于 2026-10-15已译

HACHIMI-II：从角色扮演到学习者行为建模以实现忠实的学生模拟

HACHIMI-II: From Role-Play to Learner Behavior Modeling for Faithful Student Simulation

Yilin Jiang · Fei Tan · Xuanyu Yin · Jing Leng · Aimin Zhou

中文摘要

模拟学生允许研究人员在真实学习者出现前测试辅导系统，但前提是模拟能复现特定学习者的成功与失败。常见方法让大语言模型进行角色扮演，但这未建模学习者已知内容或下一步行为，导致模型自身能力填补空白，模拟学生表现趋同且极少犯真实错误。HACHIMI-II 将学生模拟重构为对学习者下一步动作的第三人称预测。该方法基于学习者历史构建学生模型，推断当前题目或导师回合的学习者状态，并编译为行为约束以限制下一步动作。在三个项目反应数据集上，模拟与真实学生能力的 Wasserstein-1 距离比最强基线小 2.4--8.4 倍。在两个对话数据集上，HACHIMI-II 以相同知识类型、位置和机制复现真实学生错误的频率比最强基线高 1.8--5.1 倍。消融实验表明增益主要源于显式学习者状态。

关键要点

01指出传统角色扮演方法未建模学习者状态，导致模拟学生表现趋同且缺乏真实错误特征。
02提出 HACHIMI-II 框架，将学生模拟重构为基于学习者历史的状态推断与下一步动作预测。
03在三个项目反应数据集上，模拟与真实学生能力的 Wasserstein-1 距离比最强基线小 2.4--8.4 倍。
04在两个对话数据集上，复现真实学生错误的频率比最强基线高 1.8--5.1 倍，验证了方法有效性。

解读

尚无解读。

原始英文摘要

arXiv:2610.19654v1 Announce Type: new Abstract: Simulated students let researchers test tutoring systems and teaching strategies before real learners are available, but only if the simulation reproduces where a particular learner succeeds and fails. The common approach asks a large language model to role-play a student, but a role does not model what this learner knows or will do next, so the model's own competence fills the gap. Simulated students perform at a level set by the model, differ little from one another, and seldom make the real learners' errors. We model the learner's behavior and recast student simulation as third-person prediction of the learner's next action. HACHIMI-II builds a student model from the learner's history, infers the learner's state for the current item or tutor turn, and compiles it into a behavior harness that constrains the next action. On three item-response datasets, the Wasserstein-1 distance between simulated and real student ability is 2.4--8.4$\times$ smaller than that of the strongest baseline. On two dialogue datasets, HACHIMI-II reproduces real student errors with the same knowledge type, locus, and mechanism 1.8--5.1$\times$ more often than the strongest baseline, and ablations trace the gains mainly to the explicit learner state. Predicting the learner makes capable models usable as simulated students.