Uma abordagem de Game Learning Analytics para identificação de habilidades de leitura e escrita no ensino infantil

Abstract

The power that video games have to capture their players’ attention has brought with it the idea of using them with the main objective of reinforcing learning in educational context. Recent studies demonstrate that it is possible to analyze the interactions of players in such games, called Serious Games, to conclude and measure the learning obtained during interaction in those games. Given this context, this work aims to develop an analysis of data obtained from the interaction of players in one game, out of 20, applied during a research that proved their positive impact on the development of reading and writing skills of 4-years-old children. Three classifiers were selected: Naive Bayes, Support Vector Machines (SVM) and Logistic Regression, which were trained with the data resulting from the interaction of these players with the game and demonstrated the hit rate of each of the classifiers. In addition, this work also makes an analysis of the interactions considered more relevant by one of the models, finding relationships between the words proposed as challenge in the test and those present in the game, raising reflections that can be taken into account during the development of a educational game that aims to improve children’s reading and writing skills in early childhood education.