Um modelo de negociação de privacidade para sistemas de recomendação social

• Ano de defesa: 2015
• Autor(a) principal: Rocha, Ânderson Kanegae Soares
• Orientador(a): Zorzo, Sérgio Donizetti
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC
• País: BR
• Palavras-chave em Português: Privacidade e personalização; Negociação; Sistemas de recomendação; Redes sociais online; Controle de privacidade; Preocupações com privacidade; Web social
• Palavras-chave em Inglês: Privacy negotiation; Social recommender systems; Privacy control; Privacy personalization; Privacy preferences; Privacy concerns; Social web
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/602

The high rate of growth and variety of information available on the Internet can overwhelm users, not leading them to the best decisions. In this context, social recommender systems play an important role on helping users against the effects of information overload. However, these systems need for data collection from its users social context motivates privacy concerns and may discourage its use. Thus, this dissertation presents a privacy negotiation model for social recommender systems to enable user to control his own privacy from the perspective of computer science. So, the user can decide to provide access to their data considering the personalization benefits that the system can offer him in exchange and is not forced to fully accept the privacy policies though. In this model, the privacy control is possible by means of a user interface design pattern using privacy negotiation techniques. The SocialRecSys social recommender system is an implementation of this model that was used in an evaluation with 32 users. The results showed that users are not satisfied with traditional interfaces and the model can better deal with the potentially different privacy preferences of each user. The results also indicated the high usability of the user interfaces of this model, which increase the flexibility of the systems regarding the configuration options of privacy preferences without harm the usage easiness of it. The implementation of this model shows that this is an alternative to reduce the concerns of privacy of social recommender systems users by increasing the flexibility and providing them a better understanding of the recommender systems. So users can feel encouraged to share their data in social recommender systems and take advantage of its personalization benefits.