Determination of auto-aggressive behavior using machine learning methods
Friedrich Schiller Universität Jena, Olga Kanishcheva, Надія Вікторівна Бабкова, Dina Huliieva, Зоя Кочуєва, Наталія Сергіївна Угольнікова · 2024
The article is devoted to researching the possibilities of using machine learning methods to detect auto-aggressive behavior in texts, in particular, based on data from Twitter.The paper analyzed various formal and informal signs using the "Suicidal Ideation on Twitter" dataset, during which the most significant for the identification of auto-aggressive behavior were singled out.Logistic Regression and Random Forest methods were used for classification, which demonstrated satisfactory results.Further research is planned, which will include the application of neural models such as CNN, RNN (LSTM), and BERT, to compare their performance with classical methods.The obtained results indicate the prospects of using machine learning methods to detect auto-aggressive behavior in English texts, which may be extended to the Ukrainian language in the future.The obtained results can be used to improve the quality of life and reduce social exclusion for persons with a tendency to auto-aggression.