Título: Exploring Session -Based Recommender Systems with Reminders
Autores: Douglas Rorie Tanno, Vinicius Sylvestre Simm, Gabriel Libardi Lulu & Marcos Domingues
Resumo: Online services, such as e-commerce platforms, streaming services, and social networks, provide users with vast amounts of information, often leading to information overload and making it challenging to discover items of interest. Recommender systems have been developed to mitigate this issue by personalizing user experiences and facilitating the discovery of relevant content. These systems leverage user preference data, interaction history, and demographic information to generate accurate recommendations. In this work, we investigate the effectiveness of Session-Based Recommender Systems (SBRS) with reminders, which incorporate previously viewed items to improve recommendation precision and relevance. We conducted experiments in the legal, music, employment, and e-commerce domains to evaluate the generalizability of reminder strategies beyond the e-commerce setting, where they have been most widely adopted. Results show thatreminders consistently improve performance in employment and legal scenarios, corroborate the existing literature with strong gains in e-commerce, and reveal a more modest impact in the music domain, suggesting potential for further research in this area.
Palavras-chave: Recommender Systems; Reminders; Session-Based Recommender Systems.
Páginas: 8
Código DOI: 10.21528/CBIC2025-1174276
Artigo em PDF: CBIC_2025_paper1174276.pdf
Arquivo BibTeX:
CBIC_2025_1174276.bib