Generation of social network user profiles and their relationship with suicidal behaviour

Resumen

Suicide is one of the leading causes of death worldwide, so characterising individuals with such tendencies can help prevent suicide attempts. In this study, a corpus, called SuicidAttempt, of Telegram messaging app users, both with and without explicit mentions of suicide attempts, has been compiled in Spanish. For each user, different demographic features were semi-automatically annotated by different systems, some supervised and some unsupervised. Finally, the collected features and linguistic features extracted from users’ messages were analysed to characterise different groups based on their relationship with suicidal behaviour. The results indicate that by detecting these demographic and psycholinguistic features, it is possible to characterise specific at-risk groups and gain detailed insight into the profiles of those who engage in such acts.