I'm playing around with NER (Named Entity Recognition) and the basic idea is that I can pass in multiple paragraphs and get recognized entities in a nicely formatted dictionary of lists.
I might look into running the Java servelet that stanford made to increase performance.
{
'organizations': ['Wall', 'Street', 'Journal', 'Apple', 'Inc.', 'Apple', 'TV', 'Apple', 'Mac', 'App', 'Store', 'Apple', 'Computer', ',', 'Inc.', 'Apple', 'Inc.', 'National', 'Hockey', 'League', 'Montreal', 'Canadiens', 'Stanley', 'Cups', 'Toronto', 'Blue', 'Jays'],
'locations': ['France', 'Cupertino', 'California'],