Personalized live streaming channel recommendation based on most similar neighbors

摘要

With the rapid development of mobile network technology, an increasing number of viewers are watching channels through live streaming platforms, and thousands upon thousands of channels are broadcasting on the platforms as well. To create a better user environment, it is required to provide accurate channel recommendation services for viewers. The current channel recommendation method clusters viewers with similar channel preferences into the same group and gives the channel recommendation based on the preferences of all the viewers in the same group. However, viewers in the same group still have slight differences in channel preferences, the recommended channels of the method may not necessarily meet the needs of viewers. To improve the accuracy of channel recommendation, we propose a channel recommendation technique, named n-Most Similar Neighbor algorithm (n-MSN), which considers the preferences of the n viewers with most similar preferences to accurately predict the channels that might be of interest to other viewers. In the experiments, we analyze the currently popular live streaming platform, Twitch; the results confirm that the effectiveness and the efficiency of the n-MSN algorithm are better than those of the existing channel recommendation methods, and the accuracy of the n-MSN algorithm is relatively stable compared with the existing methods.