HE METHOD OF COMBINED ONLINE ASSESSMENT ENGAGEMENT IN ADAPTIVE LEARNING SYSTEMS

Abstract

Modern trends in the development of educational software are associated with adaptive learning systems that can personalize the learning process. One of the key quality indicators of any learning system is user engagement in learning process. The known methods for assessing engagement of users of computer systems are based on collecting of various data on the user’s behavior, his emotional and neurophysiological state, etc., and interpreting it in the context of involvement. Confidence in the results of assessments based on interpretations may be insufficient to make decisions based on these results, which is of fundamental importance in case of using in adaptive learning systems. And besides not all the methods can be used online, which is a prerequisite for application in adaptive learning systems. A method for online assessing engagement has been developed, based on a combination of oculography, emotion analysis and web analytics methods, which is applicable for adapting the learning process in adaptive learning systems. The joint analysis of the physiological state and behavior (actions) of the user allows to take into account various aspects of the possible manifestation of the user engagement, and increase the confidence in the results of the assessment of engagement. Quantitative engagement indicators based on metrics applicable to online engagement assessment are suggested. Due to ambiguity of possible interpretations of quantitative indicators of involvement, a generalized assessment of engagement is determined using a fuzzy inference mechanism. To assess engagement a linguistic variable is used and a method for assessing user engagement is based on fuzzy rules. The proposed method is implemented in a web application that can be used to assess involvement online in adaptive learning systems. To set the linguistic variable for assessing engagement, an experiment will be conducted with the participation of users of learning systems and external experts. The values of the linguistic variable will be defined in such a way that the result of assessing involvement, obtained using the developed method, does not contradict the results of expert assessments.