Recomendação para formação de grupos para atividades colaborativas utilizando a caracterização dos aprendizes baseada em trilhas de aprendizagem

Abstract

Groups are a basic social structure, and as inside as outside of the academic world they form and change themselves in various ways for multiple purposes. While students form groups easily out of class, forming groups on a course can be an unnatural process. However, for collaborative learning to succeed, it is important to form groups that can be effective and efficient in accomplishing the objectives of the task. In this sense, it is sought to improve the interactions of students mainly in group activities. The group work is a resource widely used by the teacher, with the intention of encouraging students' interaction in collaborative activities. There is a lack of support for group creation in Virtual Learning Environments (VLEs). In this sense, Learning Path (LPs) can be resources to suggest groups of learners. The use of LPs is promising because it shows the paths taken by students in VLEs, which makes it possible to suggest groups based on these interactions, since to perform groups in the EAD modality is really a challenging task for the teacher. In general, in this modality, students only have 20% of face-to-face contact with each other and with the teacher, making it difficult to establish criteria for grouping. In this context, this research proposes a conceptual framework for the formation of groups in collaborative activities, through data extracted from the LPs graphs, to assist the teacher in the teaching-learning process. To verify the feasibility of the proposed conceptual framework, the M-CLUSTER tool was developed that analyzes the attributes described below and suggests the formation of groups. The mechanism emphasizes the formation of groups by applying the K-Means algorithm, which is used with three similarity metrics, which are the distances: Euclidean, Manhattan and Cosine, using attributes (vertex access, quantity, dispersion and variances of the standard edges, forward and return, and student id) obtained through the data extracted from the LPs. Meetings were held with the teachers (specialists) to validate the results. In the case study, M-CLUSTER used the attributes and classified them with K-Means, obtaining three clustering results, one for each metric. The teacher chooses among the suggestions generated and makes one available to the students so they can choose their partners within the cluster, thus forming the groups. These groups suggested by the tool were validated and visualized by the teacher, from two representations, one descriptive and one visual through bubble charts. To validate the suggested groups, two activities were carried out in the discipline, in the first activity, the students chose their group partners and in the second, the groups were formed according to the suggestions of the tool. According to the results obtained from the case study shows that the tool obtained satisfactory results where 75% of students matched or improved their individual scores in relation to those achieved in the first activity.