Finding epic moments in live content through deep learning on collective decisions

Abstract Live streaming services enable the audience to interact with one another and streamer over live content. The surging popularity of platforms has created a competitive environment. To retain existing viewers attract newcomers, streamers fans often create well-condensed summary streamed However, this process is manual costly due length online events. current study identifies enjoyable moments in user-generated video content by examining audiences’ collective evaluation its epicness. We characterize what features “epic” present deep learning model extract them based on analyzing two million user-recommended clips associated chat conversations. shows that our data-driven approach can identify epic from cover various contexts (e.g., victory, funny, awkward, embarrassing). Our user further demonstrates proposed automatic performs comparably expert suggestions. discuss implications decision-driven extraction identifying diverse scalable way.