Automatic identification of account sharing for video streaming services

Abstract

According to multiple studies, account sharing is common among subscribers of video streaming services. This leads to huge revenue loss for service providers. Although they have strong financial interests to address the problem, service providers face multiple challenges when trying to identify shared accounts. On one hand, the huge volume of unstructured and noisy data makes it hard to manually process data. On the other hand, it is legitimate for family members to share an account, from anywhere and use many devices as they want. Only these accounts which are shared outside of the household are against policies. In this paper, we propose an efficient solution to address the account sharing problem. Based on a massive volume of session data, our solution builds user profile through accumulating and representing the geolocation and device usage information. Then we estimate the account sharing risk by analyzing the usage pattern of each account. The proposed solution can identify a significant number of shared accounts and help service providers to recoup a huge amount of lost revenue.