Beyond the comfort zone: digital nudges for off-profile recommendations

• Ano de defesa: 2023
• Autor(a) principal: Alves, Gabrielle Aparecida Pires
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Bias; Descoberta; Discovery; Diversidade; Diversity; Nudging; Recommender system; Sistemas de recomendação; Vies
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/55/55134/tde-05092023-161927/

In todays constantly evolving society, recommendation systems have emerged as a crucial tool in the digital age. In many application areas of recommendation systems, such as media streaming sites, the main objective of service providers is to increase user engagement by helping them discover new types of content they like. Standard collaborative filtering algorithms typically lead to some level of discovery. However, in some cases, it is useful to actively promote out-of-profile items so that users can explore new possibilities. But when showing out-of-profile items to users along with more familiar ones, new items may go unnoticed. In this research, we examine how the technique of \"digital nudging\", or subtly directing user choices in a specific direction, can help increase user attention and interest in out-of-profile content. We conducted a user study (N=1,064) in a real book recommendation application. We found that by directing users to recommended books of non-preferred genres, there was a significant increase in out-of-profile books added to reading lists, confirming the effectiveness of digital nudging. However, we also found that the subjective perception of recommendation relevance may decrease when using nudging, resulting in lower satisfaction and lower intention to reuse the system. Therefore, digital nudging in recommendations is effective in the short term, but should be used with caution, keeping an eye on users overall perceptions of quality and potential long-term harmful effects.