Once upon a time . . Back in the days when you could go into a shop with a pound and come out with a bag of monster munch, a can of coke, and still have a few pence left in your back pocket, most text mining applications relied on a technique known as bag-of-words. This technique basically took all the words in a given piece of text, disregarding their order, not caring about their context and used each words frequency to classify the entire document using predefined classes.