Uma abordagem para extração automática de learning analytics relacionadas à colaboração em fóruns educacionais

• Ano de defesa: 2018
• Autor(a) principal: FERREIRA, Máverick André Dionísio
• Orientador(a): MELLO, Rafael Ferreira Leite de
• Banca de defesa: MELLO, Rafael Ferreira Leite de, LINS, Rafael Dueire, RODRIGUES, Rodrigo Lins
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal Rural de Pernambuco
• Programa de Pós-Graduação: Programa de Pós-Graduação em Informática Aplicada
• Departamento: Departamento de Estatística e Informática
• País: Brasil
• Palavras-chave em Português: Fóruns de discussão; Learning analytics; Mineração de texto; Computação evolucionária
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/7864

The growth of distance learning in Brazil has contributed to democratizing the access to education, especially in higher education. Despite of that, physical distance can cause students a feeling of isolation, which in turn consists of one of the variables that can influence the learner to evade. To avoid such a problem, Virtual Learning Environments often encompass collaborative resources like discussion forums. The literature points that forums are highly collaborative resources because they provide a platform where the participants can debate to enrich their knowledge construction experience. However, the forums collaborative potential has not been fully exploited because most of the messages published in the discussions are from students to the instructor. Thus, it is necessary to provide methods capable of helping students to develop the ability to collaborate. For this, the instructors need to follow up the whole progress of the discussion, and this could be an enormous work as the number of posts increases. It is important to emphasize that the establishment of collaborative discussion in the forums decrease the students‘ sense of isolation, which promote the development of skills such as critical/reflective thinking. This dissertation presents an approach based on Text Mining, Machine Learning, and Evolutionary Computing, to automatically extract Learning Analytics related to collaboration in forums messages conducted in Portuguese. The proposed approach was based on a collaborative identification model, proposed by MURPHY (2004), of which five collaborative features were explored: soliciting feedback; Answer questions; Praise/Express appreciation by the other participants; Share information and resources and; Recognize the presence of the group. To evaluate the performance of the approach were conducted experiments in four databases, composed of messages from educational forums. The proposed method reached F-measure of up to 0.98. In order to measure the impacts of pedagogical mediation and the collaboration of students, a quasi-experiment was carried out in a real educational environment. The results showed that the approach provided the collaborative scenario of the forum for the mediator, enabling a formative evaluation, besides contributing to the increase of the students’ collaboration rates.