N-Screen Smart TV contents personalization system using adaptive hybrid similarity approach

In this paper, we propose N-Screen Smart TV contents personalization system. The N-Screen Smart TV provides contents to user multiple devices having different Codec, bit rate and display format at the different location and time. The proposed system considers user N-Screen device attributes and current access network speed for visual perceptual quality on the user device, and ensure that user device is capable to display with his preferences. Furthermore, we present a novel similarity method using user rating variation to improve the accuracy of the collaborative filtering. In addition, we suggest a weight-based adaptive hybrid similarity to combine collaborative and content-based filtering. The proposed system ensures that current access network is capable to stream and recommended contents have high visual quality on the user device, and improve the accuracy, cold start issue and sparsity.