Caching Videos with Category Diversity based Recommendation: Model and Algorithm
Huxin Zhang, Chunxi Li, Yongxiang Zhao, Chenyue Zhang · 2020
In recent years, adding recommendation function to video caching system has become a new research trend. However, due to the size limitation of cache space, it is difficult to obtain satisfactory results by recommending the cached videos, which may narrow the range of users' choices and thus worse their QoE. In this paper, we propose diversity recommendation based video caching, which should select videos to buffer by balancing three factors including the possible cache hit ratio, the diversity and the personalization once they are recommended, to improve user viewing experience. We build a diversity recommendation based caching model, which can flexibly support different intents of caching optimization by choosing different model parameter settings. Moreover, we formulate the video caching as an optimization problem, and then propose a heuristic video caching algorithm to iteratively select the most valuable videos to buffer, where the value of each video is calculated based on the above mentioned factors. The numerical results of the simulation based on the movielens dataset demonstrate that the proposed algorithm can greatly improve the diversity of the recommending videos at a relatively low cost of cache hit rate loss.