Análise da fidelidade pelo comportamento dos usuários do twitter

• Autor(a) principal: Matos Júnior, João Batista Pereira
• Data de Publicação: 2017
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• dARK ID: ark:/38995/0013000009tm7
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tede/7893
• topic: Influência social; Fidelidade; Personalização; Redes sociais; Twitter; Comportamento de usuário; Loyalty; Personalization; Social networks; Twitter; User behavior; Social influence; CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

Users influence analysis over other users has been highlighted in studies involving online social networks, especially for the social network known as Twitter. However, there is still room for studies which analyze the influence that a user a exerts on another user e, based on the frequency of interactions from e dedicated to a in the social network. In the case of Twitter, three interactions are commonly used by users: the like, the mention and the retweet. This work assumes that a direct way of measuring the influence of a user a over another user e on Twitter is to observe the number of interactions of e with a over time. In this sense, a new concept called loyalty is proposed to define a specific and personalized type of influence, in which users, through their interactions, demonstrate preference by some peers in the social network (i.e., through the frequency of the interactions). Three aspects of loyalty were analyzed: i) intensity of interaction with the preferred user; ii) probability of interaction with users with whom there is loyalty, and iii) predictability of user interactions with their peers. Experiments done on a large sample of Twitter users reveal that the vast majority of users have loyalty to some user. In addition, the type of interaction and cultural aspects such as the written language used in the tweets differently affect the three aspects of fidelity considered. The paper also offers suggestions on how information derived from loyalty may be useful for future work related to influence on Twitter.