Machine Learning: How to find sets of similar things

Hi all,

Does anyone have any good reading or code for finding sets of similar things? For example, given a text dataset like every comment on Reddit, how would you find lists of cities, or cocktails, or politicians? If nothing else, I want to know what the name of this problem is, so I can research it...thanks