CAREER: Looking Glass: Leveraging Mentor Interactions to Create Personalized Programming Help for Independent Learners

这个项目将扩大计算机科学家的努力，他们通过捕捉导师与学员之间的互动，并利用这些捕获的内容创建一个虚拟指导系统，以支持在Looking Glass编程环境中独立的学习者（无法接触计算机科学专家的儿童）。Looking Glass是一个新手编程环境，它将编程作为创造3D动画故事的激励手段。在待开发的系统中，导师将做三件事：首先，导师观察他或她的学员创建的故事程序，并编写代码建议：新的或修改的功能，将有助于改进学员的程序。接下来，导师编辑一个教程草案，该草案是Looking Glass根据导师的代码建议自动生成的。然后，Looking Glass将这个编辑过的教程发送给学员。最后，导师编写规则，帮助Looking Glass识别代码建议可能有用的上下文。Looking Glass将使用导师提供的代码建议、个性化教程和规则，为独立学习者提供虚拟指导。具体而言，本项目将探讨三个假设：假设一：获取和评估导师创建的学习材料的方法将使创建虚拟指导系统成为可能，以支持独立学习者。假设二：与那些没有获得这些建议的人相比，那些被提供了上下文代码建议的独立学习者，既能推进他们的故事，又能引入新的编程概念，将会发展出更好的编程技能。假设三：根据每个主题的经验水平提供个性化教程的独立学习者将比提供单一级别脚手架的独立学习者表现得更好。美国国家工程院（National Academy of Engineering）将个性化学习列为本世纪工程学面临的重大挑战之一。该项目将开发和评估两种个性化学习技术的影响：特定于程序的代码建议和个性化的多层次教程。大多数对教育环境的研究都集中在正式的教育环境上，但Looking Glass将推进如何在开放式软件环境中有效支持独立学习的研究。对Looking Glass的改进将通过虚拟辅导为没有机会学习计算机科学的中学生提供探索计算机编程和学习支持的环境。该项目的教育计划将计算机科学家的传记整合到Looking Glass和一个研究博客中，以帮助中学生培养对计算机科学的准确印象。

This project will amplify the efforts of computer scientists who do outreach to middle school students by capturing mentor-mentee interactions and using this captured content to create a virtual mentoring system to support independent learners (children without access to computer science experts) within the Looking Glass programming environment. Looking Glass is a novice programming environment that presents programming as a means to the motivating end of creating 3D animated stories. In the system to be developed, mentors will do three things: First, a mentor watches the story-programs his or her mentee created and writes a code suggestion: new or revised functionality that will help to improve the mentees' programs. Next, the mentor edits a draft tutorial that Looking Glass has automatically generated from the mentor's code suggestion. Looking Glass then sends this edited tutorial to the mentee. Finally, the mentor writes rules that help Looking Glass identify contexts in which that code suggestion could be helpful. Looking Glass will use the mentor-contributed code suggestions, personalized tutorials, and rules to provide virtual mentoring for independent learners. Specifically, this project will explore three hypotheses: Hypothesis One: The approach of capturing and evaluating mentor-created learning materials will enable the creation of a virtual mentoring system to support independent learners. Hypothesis Two: Independent learners who are presented with in-context code suggestions that both further their stories and introduce new programming concepts will develop greater programming skills than those without access to these suggestions. Hypothesis Three: Independent learners who are presented with personalized tutorials based on their experience level with each topic presented will perform better than independent learners who are presented with a single level of scaffolding. The National Academy of Engineering lists personalized learning as one of the grand challenges for engineering in this century. This project will develop and evaluate the impact of two techniques for personalizing learning: program-specific code suggestions and personalized, multi-level tutorials. The majority of research into educational environments has focused on formal educational settings, but Looking Glass will advance research in how to effectively support independent learning in open-ended software environments. The enhancement to Looking Glass will provide an environment for exploring computer programming and learning support through virtual mentoring to middle school children without access to computer science learning opportunities. The project's educational plan uses biographies of computer scientists integrated into Looking Glass and a research blog to help middle school children develop an accurate image of computer science.