Face Recognition-Aided IPTV Group Recommender with Consideration of Serendipity

The internet protocol television (IPTV) provide thousands of video contents to the users. However, it is difficult to each user to make selection from the amount of videos especially for the group of users. For dealing with the problem, the recommendation systems are developed to aid the users to make selection. In this study, the group recommender named "EyeTV" is proposed which is a multi-facial recognition technique-aided IPTV system, and it is an improved version of the 1st generation proposed in 2012. The EyeTV system applies Microsoft Xbox's IP camera - KINECT to be the eye of TV for capturing facial features of users, and the multi-facial recognition technique is employed to recognize the user's identity. The users who watch the videos together are taken as a group, and then the system would memorize the watching history for that group automatically. The interactions among the group members can be predicted by the recommendation algorithm of EyeTV, that is, the recommended results are taken the members' interactions into consideration, and the recommended results may be more and more relevant to the group after training. However, the members may be able to guess what kinds of videos would be recommended to them at the time, and the group recommender can be considered as an ineffective service. For preventing the problem mentioned before and providing the group members new watching experience, the social network-based serendipity recommender is employed to improve the EyeTV system. The improved EyeTV system takes the recorded information from the social network of each member, and the data from social network would be combined with the group's watching history for providing serendipitous recommendations to the group.