Predicting pre-knowledge on vocabulary from e-learning assignments for language learners

Abstract

In the current big data era, we have witnessed the prosperity of emerging massive open online courses, user-generated data and ubiquitous techniques. These evolving technologies and applications have significantly changed the ways for people to learn new knowledge and access information. To find users’ desired data in an effective and efficient way, it is critical to understand/model users in applications involving in such a large volume of learning resources. For instance, word learning systems can be promoted significantly in terms of learning effectiveness if the pre-knowledge on vocabulary of learners can be predicted accurately. In this research, we focus on the issue of how to model a specific group of users, i.e., language learners, in the context of e-learning systems. Specifically, we try to predict the pre-knowledge on vocabulary of learners from their previous learning documents such as writing assignments and reading essays. The experimental study on real participants shows that the proposed predicting model is very effective and can be exploited for various applications in the future.