A Multimedia Contents Recommender System for Customized Mobile Web Service

Abstract

As the mobile Web technology becomes increasingly applicable, the mobile content market, especially the character image downloading for mobile phones, has recorded remarkable growth. In spite of the rapid growth of the market, customers experience high levels of frustration in searching for the specific character images they want. The encountered problems are mainly caused by distinctive characteristics of visual contents and the mobile Web environment. We proposed a visual contents recommender system (VISCORS) to address the customers inconvenience. VISCORS combines two techniques from different research domains: collaborative filtering and content-based image retrieval. We also conducted the experiment using the real character images with participation of real mobile Web customers to evaluate the performance of VISCORS. The experiment results confirmed that the system could be used as a realistic solution for character image recommendation in the mobile Web environment.