Detecting psychological sentiments in users from social networks

• Autor(a) principal: Santos, Patrícia de Sousa dos
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/21823
• topic: Sentiment analysis; Social networks; Twitter; Reddit; Mental health; Depression; Análise de sentimentos; Rede social; Saúde mental; Depressão

Over the years there has been an increase in the population suffering from depression. Many users find refuge in social networks as an alternative to health professionals, because they feel safer and comfortable to open up with others without forcing them to express their opinion. Despite the negative impact of social networks on users, many studies show that based on the posts made, it is possible to make a distinction between depressed and non-depressed users and a behavioural and linguistic analysis. In this way, the literature reports that prediction models of depressive users have been created and improved. In this dissertation, we aim to test and develop a model that helps in the detection of depressive users in social networks based on the words and emotions of the posts, using two lexicons, NRC Emotion Lexicon and VAD Lexicon. Through the NRC Emotion Lexicon we wanted to evaluate if the implementation of new features in a study presented in the related work, allows an improvement in the performance of the model developed. With VAD Lexicon, the objective is to analyze if through VAD values it is possible to extract the emotions. We concluded that the VAD Lexicon was not advantageous since, it was not possible to correlate the VAD values with the emotions. With NRC Emotion Lexicon, we concluded that although the integration of the lexicon did not result in a better performance, it was possible to observe improvements, demonstrating that the lexicon contributed positively to the effectiveness of the model.