Intelligent Design Advisor for Web-based Instructional Games

Design advisors are the most viable method to guide novices and non-ID professionals through the process of creating effective instructional materials, and are especially useful in situations where experts are unavailable. Yet, production of accurate advisors requires tedious detailed analysis of relevant expertise, processes, and tasks. However, sufficient expertise may not always be available, or may be incomplete, inconsistent, or changing frequently. This is particularly valid for new domains, such as online instructional games. An alternative route is automated Machine Learning (ML). Author feedback and machine learning can be used to selectively and automatically refine, revise, validate, and expand the advisor's expertise, improving performance and effectiveness without incurring the repeated costs and inconvenience of manual system upgrades. We propose to undertake this promising route, with the goal of producing a web-based intelligent advisor that incorporates machine learning technology to automatically improve its effectiveness as it interacts with authors.