AutoMentor: Virtual Mentoring and Assessment in Computer Games for STEM Learning

点击翻译按钮获取中文摘要

The project builds on existing research and extends it to include a computational model of participation in a community of practice. The products include an automated mentoring system that uses natural language conversations to help students learn about science and technology and an assessment/analysis protocol to quantify students' STEM behavior. It extends automated tutoring with an automated mentoring technology, AutoMentor, and combines it with Epistemic Network Analysis. AutoMentor is implemented in Land Science, a multi-player urban planning game, and tested by the Massachusetts Audubon Society with approximately 700 middle school aged students either in after-school or in-school programs. Pioneering work drawn from three partner institutions inform the design of the system. These include intelligent tutoring (Graesser, University of Memphis), assessment (Mislevy, University of Maryland), and game-based learning (Shaffer, University of Wisconsin-Madison). The core technologies are supported by expertise in computer science, science content, measurement, and STEM educational programming. The project has the potential to transform all three core areas by providing a fully implementable system that supports student learning substantive content through immersion in a STEM processes. The PIs anticipate that this will be integral to learning systems of the future.