© 2018 ACM. This paper presents the preliminary findings of an investigation into online opinion leadership using user-generated content from one of the most popular social media digital platforms in China (Douban), where a large army of media-savvy users post and follow comments about films. We first address the problems involved in harvesting huge data samples relating to a number of top-performing South Korean films via an asynchronous scraping crawler. Then, to gain deeper insights into how opinion leaders and followers are making sense of popular cinema, especially exposure to transnational content, the data samples collected are represented using multiple high-level features. Finally, a linear regression model is introduced to analyse the characteristics of opinion leaders. These preliminary results show that the proposed framework is flexible and applicable to social media data for the mining of information relating to online leadership, information which can in turn be used to better inform domestic as well as international producers, distributors and consumers of digital media content.