Personalized mobile advertising application using Bayesian networks

Abstract

Advances in wireless technology increase the number of mobile device users and give pace to the rapid development of mobile commerce conducted with these devices. Empowered by the Web’s interactive and quick-response capabilities, mobile marketing is a very promising direct marketing channel. The mobile device provides unrivalled one-to-one marketing capabilities. However, the constraints of both mobile networks and devices present negative impact on the operational performance of mobile commerce. Consequently, personalization technologies are needed and the success of mobile commerce largely depends on whether personalization can be utilized to deliver highly personalized and context sensitive information to mobile clients. I believe that personalization plays an important role in affecting people’s attitude and intention toward mobile advertising, which is verified by my empirical study. To realize the personalization function in mobile advertising, Bayesian Network-based user-modeling technique is adopted. To build the Bayesian Network-based user model, the network structure variables and the prior probabilities are two most important fundamental elements needed to be identified at the very beginning. For this purpose, empirical study is adopted and it finds that context, content, user preferences are the important components that can be utilized to achieve personalization effect in mobile advertising application. The collected data of user information from the empirical study is used as prior probabilities for the Bayesian network as well. The personalized Bayesian Network-based prototype consists of a user model, a context model, a content component and a matching engine to deliver more relevant advertisements to customers through mobile devices. Hugin software is used to help the network reasoning and adaptation for the models and the engine. The architecture and the main interfaces of the prototype are also presented. KEYWORDS: mobile commerce, Bayesian networks, user modeling, mobile advertising, personalization, content, context, user preference, attitude, and intention