360264 keyword grouping algorithm

Here is a quick and interesting project for you algorithm lovers. You are given a set of keywords (loaded from a file). For example: banana cake recipe blue cake recipe salmon pattie recipe monkey bread recipe corned beef and bread recipe blue punch recipe jello poke cake recipe You need to output a list of keyword patterns, sorted by number of times the pattern appears in keywords: [...] recipe 7 [...] cake recipe 3 [...] bread recipe 2 blue [...] 2 The additional input parameter can be minimum number of words a pattern can have to be considered. If for example this was set at 2, the output would become: [...] cake recipe 3 [...] bread recipe 2 You should ignore noncharacter-nondigit for example "cake recipe ! " is same as cake recipe. The real trick is to write a good algorithm that will work fast - my test file contains nearly 200,000 keywords ! Sounds interesting? Apply!