Machine learning for user modeling in a multilingual learning system

Towards the successful creation of user models that can be incorporated into foreign language learning systems, we have used algorithmic approaches residing in the field of machine learning. The creation of user models is even more demanding in the area of Computer Assisted Multilanguage Learning, since modeling is an even more complex process that concurrently handles information from multiple domains. These domains have important similarities but also basic differences. This paper describes the implementation of student modeling through machine learning techniques, which aims to ameliorate future multiple language learning systems. The incorporation of k-means clustering is used to address several barriers posed by the heterogeneous learning audience. The resulting system both generates and discovers user profiles, based on students' characteristics, performance and preferences. Through our system, we promote the adaptivity and individualization to each user that interacts with the application, by providing individualized help, error diagnosis and error proneness along with advice generator components.