RECOMMENDING AND SELECTING APPROPRIATE RESOURCES DURING ON-LINE PROBLEM SOLVING
Gregory A. Krudysz, James H. McClellan · 2014
In this paper, we present a model of a personalized tutoring agent derived from a set of preliminary student data that was acquired using a web-based question and answering system. The system, called ITS, has been deployed in a second-year Signal Processing ECE course as an on-line supplement to the traditional homework. ITS allows students the opportunity to practice answering concept-centric questions and tests students’ conceptual understanding through instructor assigned questions. The long term research goal is to develop an interactive learning environment that provides personalized tutoring via a set of questions and web-based content. Through a data-driven approach, we apply Hierarchical Bayesian models to develop a probabilistic conceptual framework for establishing and tracking the conceptual state and growth of students as they interact with questions and course related resources. We present preliminary results from our system, and discuss ITS in the context of extensions to account for conceptual correlations, a priori labeling, and temporal prediction.