This is code to algorithmically generate beer reviews, based off of real reviews I scraped.
Faking beer snobbery using N-Grams and markov chains! Try it yourself by running cleanBeerReviews.py, and then generateReview.py!
Inspiration and much of the code from http://www.onthelambda.com/2014/02/20/how-to-fake-a-sophisticated-knowledge-of-wine-with-markov-chains/
Some successes:
"Light Amber color, nearly copper, medium head. Medium hopped, with some yeast haze when finishing bottle.”
“Ton of malt and sweet malts, maybe a touch of caramel. Texture is fairly bitter with a nice head. nice fruity aroma. Nicely carbonated, with a thick, camel color fine bubbled head.”
"Pours a clear coppery orange with off-white foam disapearing quickly. Lots of really nice beer that finished with a moderate amount of carbonation. Not too complex, but easy to drink, but not very dark brown with good clarity and small bubbles.”
Some fails:
“Did a Schneider tasting over the aroma of English marmalade, over-riped banana and clove dominate the aroma.”
"Not a fan - a symbol of outstanding quality and taste that's been perfected by the brewers could cram into our mash tun, and lots of lace.”
“The beer poured a huge loyal following and has a distinct fruity banana aroma. It smells more chocolate than it was, but a must when in Antwerp - although very touristic, also quality.”