Multi-Relational Factorization Models for Predicting Student Performance

Predicting student performance (PSP) is an important task in educational data mining, where we can give the students early feedbacks to help them improving their study results. A good and reliable model which accurately predicts the student performance may replace the current standardized tests, thus, reducing the pressure on teaching and learning for examinations as well as saving a lot of time and effort for both teachers and students [Feng et al. 2009; Thai-Nghe et al. 2011]. Precisely, PSP is the task where we would like to know how the students learn (e.g. generally or narrowly), how quickly or slowly they adapt to new problems or if it is possible to infer the knowledge requirements to solve the problems directly from student performance data [Corbett and Anderson 1995; Feng et al. 2009], and eventually, we would like to know whether the students perform the tasks (exercises) correctly (or with some levels of certainty). The benefits of PSP have been vastly discussed in the literature [Cen et al. 2006; Feng et al. 2009; Thai-Nghe et al. 2011]. To address the problem of PSP, several works have been published, e.g. as summarized in Romero et al. [2010], but most of them relied on traditional classification/regression techniques. For example, Corbett and Anderson [1995] proposed the Knowledge Tracing (KT) model, which is usually used for tracing the students’ knowledge in applying their skills as well as for PSP; Cen et al. [2006] proposed a semi-automated method for improving a cognitive model called Learning Factors Analysis that combines a statistical model, human expertise and a combinatorial search; Yu et al. [2010] used linear support vector machines together with feature engineering and ensembling techniques for predicting student performance. This approach, however, requires intensive computer memory and much human effort on data pre-processing. Recently, researchers have proposed using recommender system techniques, e.g. k-NN collaborative filtering and matrix factorization, for PSP [Cetintas et al. 2010; Toscher and Jahrer 2010; Thai-Nghe et al. 2011]. The literature have shown that PSP can be considered as rating prediction task in recommender systems since the student, task, and performance would become user, item, and rating, respectively. The authors also shown that matrix factorization is a promising approach for PSP.