Crowdsourced Acquisition of Models of Learning Transfer Strategies (CRAM-LESS)

The current generation of US students needs a stronger base in Science, Technology, Engineering, and Mathematics (STEM) principles, an understanding of how these principles relate to the real world, and the ability to transfer these principles to tomorrows science and engineering challenges. The Navys use of science and technology is increasing, and a lack of STEM-ready applicants could lead to a potential talent crisis. The Navy is exploring the application of intelligent tutoring systems (ITSs) to improve the efficiency and effectiveness of future training. Key to both the learning of STEM principles and the application of those principles to real world problems is the concept of learning transfer, which current ITSs do not explicitly model in a structured and theoretically founded manner. To address this need, we propose to design and demonstrate a system for Crowdsourced Acquisition of Models of Learning Transfer Strategies (CRAM-LESS). The CRAM-LESS system features (1) a theoretically grounded model of learning transfer in real-time, procedural tasks; (2) a gamified crowdsourcing methodology using both expert input and learner performance data; (3) a crowdsourcing software infrastructure that enables problem distribution and crowd data collection; and (4) probabilistic reasoning to create robust models of transfer from noisy crowdsourced data.