Raciocínio baseado em casos para apoiar a formação de grupos baseada nos traços de personalidade do modelo Big Five

Abstract

Collaborative learning comphreends the developing of knowledge through interactions among group members to achieve a shared goal. For this interaction to happen, one needs to ensure that the students grouped together will not intervene negatively in the collaboration process. Grouping carelessly can trigger undesirable scenarios, such as a student isolated in a group. For this reason, it is essential to consider a group formation strategy to promote productive interactions and to ensure that such a strategy can be used to support collaboration. Among the criteria that can be considered in group formation, the personality traits of the students indicate behavioral tendencies that may influence their performance. For example, in a group with many students with a high degree of Extroversion, they could get distracted by interaction and neglect the common goal. Thus, by knowing these characteristics one can think of developing a group formation strategy to group students and increase the occurrence of productive interactions for collaboration. In order to provide a way to detect personality traits in a transparent way and to make this information available to support group formation strategies for collaboration, this work evaluates texts written by students which can be captured in chats and forums in virtual learning environments, for example. Although the tested algorithms do not reached high accuracy in the experiments, the results pointed a chance to improve the prediction accuracy. Along with the detection of personality traits in the text, experiments with groups of students were conducted with the objective of building the case base for Case Based Reasoning. As observed in the classes in which group formation experiments were conducted, by querying the case base, the new groups formed had good interaction. Moreover, according to the teacher's report, the groups formed were perceived as better than those formed by the students themselves. Regarding the case base, the experiments conducted indicate an improvement in the quality and comprehensiveness of the cases as the base was updated. Since this update is part of the Case Based Reasoning operation process, the base will be continuously improved, as well as the quality of grouping suggestions and the information that can be extracted from it. Once the case base can grow bigger, turning the queries time-consuming, experiments aiming to extract grouping rules from the case base were also conducted. The results obtained suggest that using grouping rules is a feasible alternative to form groups, if the number of cases increases too much.