Modelo de clasificación de rasgos de adicción a los videojuegos en base a posts de Reddit mediante técnicas de minería de texto y aprendizaje de máquina.

Abstract

In recent years, a large number of studies have been focusing on the identification of a wide range of mental illnesses through the use and application of data mining, data analysis and machine learning. Currently, addiction to video games is considered a global issue that has become increasingly frequent and that directly affects the life’s quality of the people who suffer from it and also their closest family environment. In the present work, we proposed to develop a novel framework for the detection of possible traits of addiction to video games focused on a group of users inside the Reddit social network. We extracted about 987 posts written by English-speaking users with the purpose of analyzing and processing the text obtained to generate six text models using BoW, TF-IDF, Word2vec, Empath, Emolex and BERT. Those models were also evaluated and tested using four supervised classification algorithms: Logistic Regression, KNN, Decision Tree and AdaBoost. The results obtained showed that it is possible to effectively identify video game addiction traits in a group of users at risk using Word2vec by generating word embeddings over a previously trained dataset.