Aberrant Learning Achievement Detection Based on Person-fit Statistics in Personalized e-Learning Systems

A personalized e-learning service provides learning content to fit learners’ individual differences. Learning achievements are influenced by cognitive as well as non-cognitive factors such as mood, motivation, interest, and personal styles. This paper proposes the Learning Caution Indexes (LCI) to detect aberrant learning patterns. The philosophy behind the LCI is that if any non-cognitive factor influences a learner, the effect will eventually be reflected in his/her learning achievement. Therefore, it’s our explicit attempt to build a prototype system aimed at assessing aspects of learning other than cognitive factors. This study proposes a personalized elearning system based on Item Response Theory which considers both course difficulty and learner’s ability to provide adaptive learning paths. The LCI, which originates from the person-fit statistics in psychometric theory, statistically judges whether the observed learning achievement is significantly different from the achievement predicted by the Item Response Theory (IRT) models. If such an aberrant learning pattern is detected, a computer tutoring agent appears to notify and encourage that learner. Furthermore, human tutors may get involved periodically to offer further guidance to support learners with aberrant patterns. Experimental results show that such diagnostics could enhance the learning efficiency and smooth the learning experience.