Too long to read?

Here's a podcast about the highlights of our article on understanding aggressiveness on the LKML:

🎙️ www.tu-chemnitz.de/informatik/S...

💡 We need more research to understand different forms and causes of aggressiveness to automatically detect aggressiveness in written communication.
(4/4)

💡 There are multiple potential causes for human disagreement, but there is no discernible general theme.
(3/4)

➡️ Our insights:

💡Humans’ perceptions differ widely, thus aggregating human annotations is not reliable.
(2/4)