Building an Artificial Neural Network Model to Predict the Tendency of Parental Mediation Types on Internet Use by Children

Abstract

Internet use by children is increasing during the Covid-19 pandemic. Therefore, parental mediation is needed to minimize the negative impact of internet use by children. This study aims to create a model based on an artificial neural network (ANN) to determine the relationship between factors in the family with parental mediation techniques in the Bogor area. The NN learning method used in this study is the Backpropagation learning method. We include the following factors in the study as the inputs of the NN, i.e. the age of parents, education, number of children, age of children, duration of using the internet, and number of social media accounts used. The types of parental mediation used as network outputs are active mediation of general internet use, active mediation of shared use, passive mediation of shared use, mediation of restrictions on internet activities, mediation of restrictions on general internet use, active mediation of internet security, monitoring mediation, and technical mediation of internet usage. We obtained the research data through a survey of 282 parents in the Bogor area in February-June 2021. This study has built an ANN model to predict the tendency of parental mediation types with a mean-squared error of 0.05132. The resulting model can be further developed into a simple educational application that can be used by parents to find out what type of mediation they are doing. By better understanding the types of mediation they do, we hope that parents can have a better understanding of parental mediation and can apply mediation techniques that are most appropriate to the conditions they experience to create family resilience.