Um modelo baseado em trilhas de aprendizagem para a representação de alunos de ambientes virtuais de aprendizagem

Abstract

Personalized educational systems, which seek to adapt the learning environment to the needs and profiles of students, generally need to use user data, machine learning and data mining, recommendation algorithms, and cataloged resources/activities. Analytical tools and models that allow the understanding of students' behavior can infer individual or collective patterns and improve students' experiences. Such tools and models can also help teachers and tutors to monitor students' actions in a Virtual Learning Environment (VLE). In this context, it is considered that new information from the VLE data can be generated, even if they do not necessarily represent the student's knowledge level, they can help facilitate the teaching-learning process. The VLEs collect data about users that can be useful to define the student's profile, his/her behavior and to identify his/her difficulties and needs. One way to accompany students is to observe the actions they take in the environment, and these actions can result in paths known as Learning Paths (LPs). This work describes a model based on Learning Paths for the representation of students from Virtual Learning Environments. The proposed model is based on the students' actions while using a virtual learning environment, the model can represent the learning paths and assist in identifying the motivational profile of each student. A quasi-systematic review was conducted in order to identify how the LPs are represented and whether there were representations for teachers or students. Research on user behavior in the educational context was also conducted. The learning paths were used to identify the behavior, specifically the goals orientation/motivation, an approach based on the Theory of Achievement Goals, developed in the area of educational psychology. The model was used to analyze the relationship between the learning paths and the students' goal orientations. To collect students' goal orientations, an instrument validated by researchers in the field of educational psychology called the Escala de Motivação de Aprendizes Universitários (Scale of Motivation for University Learners) was used. Based on previous studies and the participation of two psychologists, a taxonomy was developed for the classification of learning strategies and a set of learning strategies classified according to the proposed taxonomy. Learning strategies were extracted from scales of learning strategies validated in the literature. Based on the literature, classes of taxonomy strategies were associated with each goal orientation. Thus, students received suggestions for learning strategies according to their motivation guidelines. In the learning strategies recommendation experiments, an average rating of 4.3 was obtained on a 5-point Likert scale. In the goal orientations classification experiments, based on the analysis of the learning paths, an F1-Score above 80% was obtained, with less than a month from the beginning of classes, for tree-based classifiers.