Jingwen Zheng
  • Jun 28, 2020

Youtube video's comments analysis Jay Chou Mojito

I extract one video's (Mojito Jay Chou) comments on YouTube with YouTube Data API and in this blog I will analyse them in different views.

Jay Chou is a Taiwanese musician, singer, songwriter, record producer, film producer, actor, and director. In 2000, Chou released his debut album, titled Jay (2000), under the record company Alfa Music. In 2001, Chou released his second album, titled Fantasy (2001), which combined Western and Eastern music style. Since then his music has gained recognition throughout Asia, most notably in regions such as Mainland China, Hong Kong, Taiwan, Singapore, Southeast Asia, Korea, Japan and in the Asian communities of Western countries such as the United States and Australia. I’ve listened to his songs for nearly 20 years. Recently, he released his new song “Mojito” on YouTube, which attracts me to get a better understanding of the audience’s opinions. Thus, I extract its comments on YouTube with YouTube Data API and in this blog I will analyse them with the following points:

  • Daily comments count
  • Comments during 24 hours after the release
  • Word clouds

Daily comments count

daily-comments-count

Until 22h37 28 June 2020(GMT+2), there are 14817539 views, 16007 comments, 164641 likes and 7256 dislikes on the video on YouTube. This plot describes the daily comments count since the release. The attention that the video is released at 16h on 11 June 2020 (Central European Summer Time, GMT+2), which means there are nearly 3500 comments during 8 hours after release !! Afterward, the comments count decreases with time, but around 100 comments per day.

Comments during 24 hours after the release

first24h-comments-count

Here I zoom the first 24 hours since the release, show the comment counts of every 5 minutes with this graph. The first comment is posted only 12 seconds after release! There are 383 comments during the first 5 minutes since release. The amount decreases with time, but it has around 15 - 20 comments per 5 minutes after 18h on 11 June 2020.

Word clouds

comments-all

It might be a good idea to show comments with word clouds. Before creating word clouds, I used jieba module and nltk module to tokenize comments in Chinese and non-Chinese to split off punctuation other than periods; then created word clouds with wordcloud module.

Comments in Chinese

comments-cn

Most of Chinese-written listeners like this song, especially its prelude and the rap in the middle part, they also think the prelude is similar to one of his old song “Mi Die Xiang”(Rosemary).

Comments in non-Chinese

comments-not-cn

Non-Chinese-written listeners love this song as well, especially its Cuban style.

If you are curious about the scripts, you will find them here.

Reference

Search