How the classifier on Stanford NER works?

Question

I try to use Stanford NER to classify some text contain Call for Papers information. I have some question on it.

1. how Stanford NER works on using classifier? how the label can assign to an entity and how the tool identify an entity?
2. I have an example like there "San Francisco, USA".. Why San Francisco labeled as LOCATION, but USA labeled as ORGANIZATION? is there specific reason?
3. And last but not least, the classifier is it include the dictionary of words or how?

Thank you very much.