Bayesian models for student and user modeling

个性化和适应性界面依赖于用户需求、兴趣、目标和技能的模型。个性化学习指南是最近出现的一些创新应用程序的成功范例。它们引导用户完成学习过程，为考试、文凭、新工作或简单地自我调整进度、自我激励的学习议程做好充分准备。他们充分利用计算机可以提供的一对一、个性化辅导，方法是将用户的精力仅集中在最相关和需要进一步学习的主题上。一对一培训显示，在课堂环境下，学生的表现提高了两个标准差。认知技能评估的用户模型是这种个性化、适应性应用程序的关键组件。除了学习指南和学习环境，它们也是适应智能帮助和助手中的解释，或适应自适应超文本和自适应应用程序中的细节和词汇水平的基本组件。当前研究计划的主要目标是开发从测试数据自动建立的技能评估的概率模型。我们的目标是从小数据集构建技能评估模型，但具有细粒度的技能表示。我们倡导的方法已经被证明在预测准确性方面优于当前的方法。该方法除了在计算上高效和完全自动化外，还可以导致通用的、易于使用的、以及用于许多高级和智能接口的经济高效的工具。

Personalized and adaptive interfaces rely on models of the user's needs, interest, goals, and skills.  Personnalized study guides are successful examples of some innovative applications that have recently emerged.  They guide the user through a study process that can lead to an adequate preparation for an exam, a diploma, a new job, or simply a self-paced, self-motivated learning agenda.  They take full advantage of the one-on-one, personalized tutoring that a computer can offer by focusing the user's effort solely on the topics that are most relevant and require further study. One-on-one training was shown to improve student performance by a factor of two standard deviations over a classroom setting.User models for cognitive skills assessment are critical components for such personalized, adaptive applications. Besides study guides and learning environments, they are also fundamental components for adapting explanations in intelligent help and assistants, or adapting the level of details and vocabulary in adaptive hypertext and adaptive applications in general.The current research program's primary objective is to develop probabilistic models for skills assessment that are automatically built from test data.  We aim to construct skills assessment models from small data sets, yet with fine grained representations of skills.  The approach we advocate was already shown to outperform current approaches in terms of predictive accuracy.  In addition to being  computationally efficient and entirely automated, the approach can lead to generic, easy to use, and cost effective tools for a number of advanced and intelligent interfaces.