Mobile data services present some unique opportunities for advertisement that other media platforms (print, TV, PC Internet) have not offered before, namely the ability to associate any mobile media session with a unique mobile user, and the user's location. Hence, there is an opportunity to understand the preferences of a mobile subscriber from the mobile media sessions the user chooses to engage in over days and weeks, as opposed to a single session (like a search or a website visited) possible for PC-based Internet access, and select and send the most useful and appealing advertisement content, in a context (time/location) the mobile user would benefit most.

In this research, we propose to develop techniques to identify a mobile user's preferences from his/her mobile data usage (in particular the growing trend of watching videos from mobile devices), and select and place advertisements in a personalized, interactive, and context aware manner. At the heart of our approach is understanding the different "categories" of advertisements, and defining "user preference categories" based on the advertisement categories available. Consequently, for each mobile user engaging in a video session, we will identify the category of the video watched, using novel video classification/categorization techniques proposed. We will develop a scalable technique to identify the top user preference categories for a mobile user, based on the categories of the videos watched so far, while ensuring the privacy of the user (no data stored - only preference statistics). We will also develop techniques to select ads, taking into consideration short term, long term, and geographical preferences of a user, as well as the location of the user. Finally, we will develop novel interactive advertisement techniques, which increase the level of subscriber engagement while displaying the advertisement.

back to top





People
 

	Sujit Dey	PI
	Truong Nguyen	Co-PI
	Bill Lin	Co-PI
	Nuno Vasconcelos	Co-PI

back to top





Sponsors

back to top