Recommending highlights in Anime movies: Mining the real-time user comments “DanMaKu”

摘要

DanMaKu is the real-time interactive comments showing overlay on a web video. Such DanMaKu websites are popular in Asia, especially in Japan and China. In this paper, an analysis of nine Japanese Anime movies based on DanMaKu counts will be presented. Highlight moments and key sentences of each movie are selected based on the analysis of DanMaKu counts. An experiment is designed to test the effect of the selected highlight moments and key sentences.