bunsenfeng / twibot-20 Goto Github PK

Hi,

Thanks for releasing this larger and more comprehensive dataset. I was wondering if there was a version that also contains the full property data about the tweets themselves, and if it would be possible to gain access to that as well.

Thanks

Please tell me how the processed data in the data is processed. Can you provide specific processing steps or processing codes?

How are each user's extended followers and followings chosen?
According to your description, the first two layers are marked data sets. According to the marked data, I find that the proportion of bots is higher than that of humans.
However, according to the statistics of some papers, the proportion of robots on Twitter is about 15%, but the proportion of bots on Twibot-20 is much higher than 15%. Is this related to the expanded 10 followers and followings?

前辈好，我是一名在读大二本科生，看了您在tweet机器人识别问题上不断取得的成就，在充满崇敬的同时也倍受鼓舞。我目前在研究自然语言处理方面的问题，并试图更好地解决社交媒体机器人识别问题，因此希望获取twibot-20 dataset，不知能否获取您的同意。

Why there are isolated nodes in the dataset? As described in the paper, the data collected through breadth-first search and user attention relationship should not contain isolated nodes, that is, there must be other nodes connected to the node.