Learning to Predict Ads Click Based on Boosted Collaborative Filtering

摘要

This paper addresses the topic of social advertising,which refers to the allocation of ads based on individual usersocial information and behaviors. As social network services(e.g., Facebook and Morgenstern) are becoming the mainplatform for social activities, more than 20% of onlineadvertisements appear on social network sites. The allocationof advertisements based on both individual information andsocial relationships is becoming ever more important. In thisstudy, we first propose the notion of social filtering andcompare it with content-based filtering and collaborativefiltering for advertisement allocation in a social network.Second, we apply content-boosted and social-boosted methodsto enhance existing collaborating filtering models. Finally, aneffective learning-based framework is proposed to combinefiltering models to improve social advertising. The experimentsare conducted based on datasets collected from a social financeweb site called Morgenstern. We performed a series ofcomparison experiments between filtering approaches. Theexperimental results indicate that the learning-basedframework is able to achieve better performance results thanfundamental filtering and boosted filtering mechanisms alone.