Do your friends make you buy this brand?

• Autores: Minh-Duc Luu, Ee-Peng Lim
• Localización: Data mining and knowledge discovery, ISSN 1384-5810, Vol. 32, Nº 2, 2018, págs. 287-319
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Consumer behavior and marketing research have shown that brand has significant influence on product reviews and product purchase decisions. However, there is very little work on incorporating brand related factors into product recommender systems. Meanwhile, the similarity in brand preference between a user and other socially connected users also affects her adoption decisions. To integrate seamlessly the individual and social brand related factors into the recommendation process, we propose a novel model called Social Brand–Item–Topic(SocBIT). As the original SocBIT model does not enforce non-negativity, which poses some difficulty in result interpretation, we also propose a non-negative version, called SocBIT$$\varvec{^+}$$ + . Both SocBIT and $$\hbox {SocBIT}^+$$ SocBIT+ return not only user topic interest, but also brand-related user factors, namely user brand preferenceand user brand-consciousness. The former refers to user preference for each brand, the latter refers to the extent to which a user relies on brand to make her adoption decisions. Our experiments on real-world datasets demonstrate that SocBIT and $$\hbox {SocBIT}^+$$ SocBIT+ significantly improve rating prediction accuracy over state-of-the-art models such as Social RegularizationMa et al. (in: ACM conference on web search and data mining (WSDM), 2011), Recommendation by Social Trust EnsembleMa et al. (in: ACM conference on research and development in information retrieval (SIGIR), 2009a) and Social RecommendationMa et al. (in: ACM conference on information and knowledge management (CIKM), 2008), which incorporate only the social factors. Specifically, both SocBIT and $$\hbox {SocBIT}^+$$ SocBIT+ offer an improvement of at least 22% over these state-of-the-art models in rating prediction for various real-world datasets. Last but not least, our models also outperform the mentioned models in adoption prediction, e.g., they provide higher precision-at-Nand recall-at-N.