A recommendation mechanism for contextualized mobile advertising

摘要

Mobile advertising complements the Internet and interactive television advertising and makes it possible for advertisers to create tailor-made campaigns targeting users according to where they are, their needs of the moment and the devices they are using (i.e. contextualized mobile advertising). Therefore, it is necessary that a fully personalized mobile advertising infrastructure be made. In this paper, we present such a personalized contextualized mobile advertising infrastructure for the advertisement of commercial/non-commercial activities. We name this infrastructure MALCR, in which the primary ingredient is a recommendation mechanism that is supported by the following concepts: (1) minimize users' inputs (a typical interaction metaphor for mobile devices) for implicit browsing behaviors to be best utilized; (2) implicit browsing behaviors are then analyzed with a view to understanding the users' interests in the values of features of advertisements; (3) having understood the users' interests, Mobile Ads relevant to a designated location are subsequently scored and ranked; (4) Top-N scored advertisements are recommended. The recommendation mechanism is novel in its combination of two-level Neural Network learning, Neural Network sensitivity analysis, and attribute-based filtering. This recommendation mechanism is also justified (by thorough evaluations) to show its ability in furnishing effective personalized contextualized mobile advertising.