SBIR Phase I: Developing a Cost-Effective Method for Creating Cognitive Models for Cognitive Tutors

这个小企业创新研究（SBIR）第一阶段项目解决了编写智能辅导系统的困难。事实证明，智能辅导系统在提供基于计算机的教学方面非常有效，但从历史上看，它既昂贵又难以构建，需要人工智能和生产系统编程方面的专业技能。该方案描述了一个软件开发工具包（SDK），该工具包由四个部分组成：认知模型创作、问题创作、工具创作和课程创作。建议的研究活动集中在这些组成部分中的第一个：认知模型创作。认知模型创作由三个独立的步骤组成：定义对象层次结构，定义问题任务的目标结构，以及表示教学系统的行为。建议是定义这些步骤的面向对象可视化，以便非认知科学家可以创建认知导师。该工具将减少为有经验的认知建模者编写导师的认知模型部分所需的时间，并且它还将减少没有认知建模经验的新人创建认知模型所需的时间。认知导师SDK将产生两个影响：1)更容易生产新的认知导师，从而能够更快地将它们推向市场；2)开发一个软件开发工具包，可以独立销售，使其他公司能够在其他领域、语言、国家和市场生产智能导师。Cognitive Tutor SDK有四个市场：内部开发者、对相关市场（监狱、福利-工作计划）改编现有导师材料感兴趣的外部用户、相关数学学科的内部和外部导师开发者以及其他领域的外部导师开发者。对教育问责制的强调为有效教学提供了强大的市场需求，这应该有助于推动对智能辅导系统等成熟技术的需求。

This Small Business Innovation Research (SBIR) Phase I project addresses the difficulties of authoring intelligent tutoring systems. Intelligent Tutoring Systems have proven to be highly effective in delivering computer-based instruction, but have historically been expensive and difficult to build, requiring specialized skill in Artificial Intelligence and production systems programming. This proposal describes a Software Development Kit (SDK) composed of four components: Cognitive Model Authoring, Problem Authoring, Tool Authoring, and Curriculum Authoring. The proposed research activity centers on the first of these components: Cognitive Model Authoring. Cognitive Model Authoring is comprised of three separate steps: defining an object hierarchy, defining the goal structure of the problem task, and representing the behavior of the instructional system. The proposal is to define an object-oriented visualization of these steps, so that non-cognitive scientists can create cognitive tutors. This tool will decrease the amount of time it takes to author the cognitive model portion of a tutor for an experienced cognitive modeler, and it also will decrease the amount of time it takes for a new person with no cognitive modeling experience to come up to speed with creating cognitive modelsThe Cognitive Tutor SDK will have two impacts: 1) easier production of new Cognitive Tutors, and hence the ability to bring them to market more quickly and 2) development of a Software Development Kit that could be independently marketed, enabling other companies to produce intelligent tutors in other domains, languages, countries and markets. There are four markets for the Cognitive Tutor SDK: internal developers, external users interested in adapting existing tutor materials for related markets (prisons, welfare-to-work programs), internal and external developers of tutors in related mathematical disciplines, and external developers of tutors in other domains. The emphasis on accountability in education provides a strong market need for effective instruction, which should help drive the desire for proven technology like Intelligent Tutoring Systems.