Recommender system for learning objects based in the fusion of social signals, interests, and preferences of learner users in ubiquitous e-learning systems

摘要

In this paper, we present a recommendation approach for learning objects (LOs) in ubiquitous e-learning systems. Many of these systems are social learning networks, and learners can interact with other users through forums or chats. In these systems, learners usually perform a set of choices or make decisions (what to learn, how to learn, with whom to learn, among others) during learning, depending on the system. The developed approach uses the result of these choices as a source of information. It is an extension of the User-based Nearest Neighbor recommendation approach, which has roots in the Nearest Neighbor search problem. Moreover, this approach uses social signals, interests, and preferences of learner users. With the fusion of these elements, we sought to find the most similar users to the active user, and then, to generate more accurate recommendations. We present an experimental evaluation of this approach showing that the usage prediction accuracy varies according to the combination of user choices and presents statistically significant higher prediction than baseline approaches. Despite being focused on ubiquitous e-learning systems, we briefly discuss how to use it in other domains where we observe that users can make decisions when interacting with other systems.