Classificação de sentimentos em texto com definição de perfis de usuários móveis para redes de próxima geração

• Ano de defesa: 2020
• Autor(a) principal: Morais, Leonardo Pereira de
• Orientador(a): Borges, Vinicius da Cunha Martins
• Banca de defesa: Borges, Vinicius da Cunha Martins, Silva, Nádia Félix Felipe da, Rosa, Thierson Couto, Immich, Roger Kreutz
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Programa de Pós-graduação em Ciência da Computação (INF)
• Departamento: Instituto de Informática - INF (RG)
• País: Brasil
• Palavras-chave em Português: Redes 5G; Análise de sentimentos; Perfis de usuários; Redes sem fio de próxima geração; Processamento de linguagem natural
• Palavras-chave em Inglês: 5G networks; Sentiment analysis; User profiles; Next generation wireless networks; Natural language processing
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/10501

The accelerated growth of applications and scenarios with different requirements and characteristics, has leveraged the demand for a mobile Internet service with better performance in the next generation of wireless and mobile networks (5G and 6G networks), at the same time that the network resources like spectrum does not increase in the same proportion. Therefore, achieving quality and good experience for users of futuristic mobile networks becomes a challenging task. Technologies of softwarization aim to flexible the wireless networks and transformed them into more versatile networks, helping to minimize the challenges of the next generations. The study of user behavior enables defining profiles of mobile users, allowing take full advantage the versatility for next generation wireless networks. The influence of feeling in the use of wireless and mobile networks is few investigated in definition of the profile of the mobile user. This work aims to define profiles of wireless users of the next generation using sentiment classification in texts from WhatsApp using deep neural networks and Lexicon approach for this task. Results presented in Lexicon of higher feelings as RNP in the task of classification of feelings in text, obtaining as accuracy values for positive and negative classes of 78% and 72%, respectively.As the result of the Association Rules for the first proposed scenario presents inconclusive results. In the second scenario, the results show a strong relationship between feeling and data consumption, a relationship present in 51.8% of users. Thus, to characterize a profile that has a tendency towards high data consumption with a positive feeling with a confidence of 83%.