Real-Time Recommendation System for Online Broadcasting Advertisement

In online broadcasting, users are exposed to advertisements for various items. Traditional advertising systems do not satisfy the expectations of various users because they provide advertisements without considering the characteristics of individuals. Personalized advertisement services can be provided by introducing recommendation algorithms that take account of users’ context and history. However, since the existing recommendation system is based on users’ consumption history, it does not quickly reflect the users’ interests that change according to items appearing in the content. In addition, when the user’s history is sparse, the performance of the recommendation system is degraded. In this paper, we propose a recommendation system for online broadcasting advertisements. The proposed system calculates the similarity between users based on the user’s region of interest (ROI). The user’s preference for the item is predicted by comparing the rating history of similar users. To reduce the time for calculating the similarity between users, a tree-structured user profile model is introduced. Finally, we conduct experiments to evaluate the performance of the proposed advertisement recommendation system.