It looks like the biggest issue is guessing locations. I have few ideas how to improve it, although fixing it may be hard. Number of possible formats is huge, and event then the same format could be two different things eg. multiple locations (London, Berlin) vs location with state (San Francisco, CA)
I was trying fix most of the places posted in comments here.
For those asking how I map locations. I'm using text tagging with named entity recognition approach.
It looks like the biggest issue is guessing locations. I have few ideas how to improve it, although fixing it may be hard. Number of possible formats is huge, and event then the same format could be two different things eg. multiple locations (London, Berlin) vs location with state (San Francisco, CA)
I was trying fix most of the places posted in comments here.
For those asking how I map locations. I'm using text tagging with named entity recognition approach.