A carregar...
Publicação
Presentation Bias in movie recommendation algorithms
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.88 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
The emergence of video on demand (VOD) has transformed the way the content finds its audience.
Several improvements have been made on algorithms to provide better movie recommendations to individuals. Given the huge variety of elements that characterize a film (such as casting, genre, soundtrack, amongst others artistic and technical aspects) and that characterize individuals, most of the improvements relied on accomplishing those characteristics to do a better job regarding matching potential clients to each product. However, little attention has been given to evaluate how the algorithms’ result selection are affected by presentation bias. Understanding bias is key to choosing which algorithms will be used by the companies. The existence of a system with presentation bias and feedback loop is already a problem stated by Netflix. In this sense, this research will fill that gap providing a comparative analysis of the bias of the major movie
recommendation algorithms.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Statistics and Information Management, specialization Information Analysis and Management
Palavras-chave
Recommendation system Algorithm Presentation bias Netflix Video on demand (VOD)