Towards understanding learning behavior patterns in social adaptive personalized e-learning systems.

Abstract

Implicit user modeling has always long since played an important role in supporting personalized web-based e-learning environments and is increasingly important in other learning environments such as serious games. Its main concern is to unobtrusively and ubiquitously learn from a learner’s previous experiences and characteristics, in order to adapt the services to their personal needs. An empirical investigation for understanding learning behavior patterns forms the basis for establishing stronger implicit user modeling mechanisms and this study aims to get a better insight into types of learning behavior. The proposed usage of data mining and visualization elicited some interesting learning behavior patterns. We analyzed these from two perspectives: action frequency and action sequences, based on an expert-designed classification of behavior patterns that helped rank the various action categories according to significance from a user’s perspective. The initial results of the study are promising and suggest possible directions for further improving implicit user modeling