A context-aware recommendation system for improving the performance of targeted mobile advertising

Mobile advertising has evolved into an important category of interactive advertising because it enables advertisers to target users considering contextual factors (location, activities, devices etc.). Logically, this makes mobile applications better advertising platform to distribute advertisement enhanced by recommendation system. A recommendation system can efficiently suggest the most appropriate content of interest to users according to their preferences. Few prior studies have tried to incorporate context-awareness into the recommendation system particularly in domain of mobile advertising. A key challenge is complexity of mobile contextual information and scalability of required algorithms. This study presents context-aware collaborative filtering algorithms improving the relevancy of the prediction results. We first define context-awareness of mobile advertising scenario, and then apply the context similarity to measure a novel user-context - advertisement model with tensor factorization. We propose an algorithmic extension of multiple-dimensional collaborative filtering to show that our proposed system can outperform to this problem.