There are a lot of research going on tweets like fake news, event detection, sarcasm detection etc., So depending upon situation, people may leverage a common observable pattern to filter relevant tweets from the tweet stream.
Have anyone come across any such implementation or research?
How such application can be proved to be useful from theoretical/implementation/research view point, apart from the fact they will be time efficient?