Detection of Health-Preserving Behavior Among VK.com Users Based on the Analysis of Graphic, Text and Numerical Data

Health preservation is one of the urgent priorities for any group of people. There is a lot of research currently underway on diagnosing and monitoring health using data from social media. In this paper, the problem of the automatic classification of users of the Russian-language social network VK.com in terms of whether they lead a healthy lifestyle is considered. To solve this problem, various types of information was collected from user profiles: text, numerical and graphic data. The users then took a lifestyle and health survey. The results of this survey were used in order to split the users into groups according to the degree of adherence to a healthy lifestyle. The survey results were used to train various binary classifiers. The best results (about 0.76 F1-score) in our experiment were shown by a model that was trained on combined features (images from users public “walls”, as well as N-gram features compiled from text from the users public “walls”). These results were achieved using the following machine learning models “multilayer perceptron”, “naive Bayesian classifier” and “k nearest neighbours”.