Online videos are widely used to share content for a variety of entertainment, educational and other purposes. To support social interaction, several video-sharing websites - including Ustream, niconico, and Twitch - allow users to post messages while they are watching videos. As users' comments can be sorted according to the timecode of each video, this is known as time-anchored commenting. We propose a novel visualization method, ToPIN, which is able to analyze and categorize the topics and content types of users' time-anchored comments. We have also developed a visualization interface that combines the visualization techniques of ToPIN and ThemeRiverto generate additional valuable insights for analysts seeking to make sense of time-anchored comments. To test the utility of our approach, we visualized time-anchored commenting data from two online course videos and invited the course instructors to evaluate our system.