Um framework para identificação e monitoramento de perfis e comportamentos de consumidores baseado no uso de aplicativos em dispositivos móveis

• Ano de defesa: 2019
• Autor(a) principal: Machado, Nielsen Luiz Rechia
• Orientador(a): Ruiz, Duncan Dubugras Alcoba
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação
• Departamento: Escola Politécnica
• País: Brasil
• Palavras-chave em Português: Identificação de Perfis; Monitoramento de Perfis; Monitoramento de Comportamentos; Aprendizado de Máquina; Aplicativos Móveis
• Palavras-chave em Inglês: Profile Identification; Profile Monitoring; Behavior Monitoring; Machine Learning; Mobile Apps
• Área do conhecimento CNPq: CIENCIA DA COMPUTACAO::TEORIA DA COMPUTACAO
• Link de acesso: http://tede2.pucrs.br/tede2/handle/tede/8781

It is possible to observe a significant growth in the use of mobile devices as well as the use of applications on such devices over the last years. In addition, the technological innovation and fierce dispute to conquer the market make mobile device manufacturers companies increase their attention to the interests of their clients. These clients perform daily many activities through the use of applications, which generates, in real time, a large number of events. Therefore, it is important for aforementioned companies to understand how their customers use applications on their devices. In this sense, automatic mechanisms, capable of assisting in the identification and monitoring of profiles and behavior of such clients, can contribute to the decision making of the stackholders. Based on this, this study proposes a framework for the identification and monitoring of the profiles and behaviors of app usage on mobile devices. To achieve this goal, Data Mining techniques such as Transformation and Discretization, Machine Learning tasks such as Association Rules and Clustering, and Novelty Detection techniques such as Concept Drift and Concept Evolution, are used to explore the app usage, identify app usage patterns, pinpoint profiles, and monitor customer behaviors over time. In order to make a comparative analysis, we have evaluated the approaches adopted by the literature, considering a real app usage data stream. Results of the experimental analysis show that the proposed framework presents better results to the addressed scenario pointing to profiles and behaviors that evolve according to the data stream.