Often there is no difference between war crime evidence and war crime glorification that machine learning could discern. Exactly the same content could be inter

Often there is no difference between war crime evidence and war crime glorification that machine learning could discern. Exactly the same content could be interpreted as "look at us do great things in defense of our noble ideals!" and "look at these monsters do horrific things for no justifiable reason!". The difference is in the audience's mindset - which is only partially influenced by the uploader's intentions, and partially by how other pages and channels link to the video and present it, and partially by historical context (the same content can acquire a different interpretation five years down the road). Machine learning cannot be expected to emulate that.