Exploiting Semi-supervised Learning in the Education Field: A Critical Survey

Educational Data Mining and Learning Analytics are two interrelated and fast-growing research fields with a view to extracting meaningful information from educational data and enhancing the quality of learning. Predicting student learning outcomes is one of the most significant problems facing these fields. Addressing effectively a predictive problem comprises the training of a supervised learning algorithm on a given set of labeled data. The difficulty of obtaining a sufficient amount of labeled data in many practical problems has resulted to the development of new machine learning approaches which are generally referred to as Weakly Supervised Learning. Semi-Supervised Learning and Active Learning constitute the main components of Weakly Supervised Learning with a view to exploiting a small pool of labeled examples together with a large pool of unlabeled ones in the best possible manner for building highly accurate and robust learning models. Over the last few years, a plethora of Semi Supervised Learning algorithms have been developed and implemented with great success for solving a variety of problems in many scientific fields, among which the education field as well. Following up on recent research, the main purpose of the present study is to provide a comprehensive review on the applications of Semi Supervised Learning in the fields of Educational Data Mining and Learning Analytics. The analysis of the relevant studies reveals that Semi Supervised Learning constitutes a very effective tool for both early and accurate prognosis of student learning outcomes anticipating better results from traditional supervised methods.