The dataset is created through the Bilibili API for the top 900+ videos under the search keyword "蔡徐坤" as of 11/25/2022

The dataset is separated into 4 folders: kun_comments, kun_complete, kun_danmakus, kun_utils

• kun_comments (the comments from each of the videos in the format of {aid}.csv)
• kun_danmakus (he danmakus comments for each video in the format of {oid}.csv)
• kun_utlls
  • aid.csv: video title with its corresponding aid
  • bvid.csv: video title with its corresponding bvid
  • oid.csv video title with its corresponding oid
  • kun_search.csv: video search result of the top 999 videos
• kun_completes (all the comments and danmakus)
  • kun_comment_parquet: all the comment in parquet format
  • top_#_kun_comment_parquet: the comment of the top # videos in parquet format
  • kun_comment.csv: all the comments in csv format
  • top#_kun_comment.csv: the comments from the top # videos

The code will release sometime in July!