www.nthba.org

I know this is weird but i need beer recipes that are bad like early experiments that weren't turning out right or recipes that are just bad.

I know this is a weird bit to ask for...
But why I'm looking for these is for a little project I'm working on.

To start off I'm a Cognitive Science major at the University of Texas at Dallas; to generalize the field of study we create artificial neural net models. These are typically discrete dynamic systems that learn; learning machines.

As a brewer my self I've always been interested in trying new things with beer recipes.

So why not combine my field of study and brewing?

This said, the project is a beer creating learning machine that uses an array of Jordan sequential neural nets as a fitness judging system for a diminishing neural net genetic model. Ok enough Jargon.

To train the neural net models I need good beer recipes that are highly rated along with mediocre and just plain bad recipes.

The thing with recipes good and mediocre ones are easy to get they are posted all over the internet. Bad ones not so much.

I'll take good recipes too but, i need bad ones more.

The beer recipes need to be rated on a scale of 1 to 10.

They need to bad due to processes and portioned ingredients; botching a good recipe with cross contamination won't help. So, to the best of your ability describe why you don't like the beer recipe. The more data the better.

Rewarding the best part;
1. you'll have my ever lasting gratitude.
2. You can have a free copy of the code for personal use assuming you have a copy of Matlab.
3. If given the time I will try rewrite the beer judging code in java along with that I will donate a working applet to the NTHBA (no guarantee on this one)
4. Contributing to science is awesome.
5. Recognition and reserved rights to your beer recipes.

Funny request. Who'd admit having a "bad recipe?"

Seriously, consider beer styles. A 10 out of 10 Helles Bock is a bad Traditional Bock, an even worse Octoberfest, a really bad Pilsner, and a terrible Barleywine. I think I just increased the difficulty of this project by orders of magnitude.

Just kidding. Good luck. Keep us updated.

Yes, I've decided to adjust the methodology and use a latent semantic analysis instead based on a give beer style.

I'm still trying to figure out a which beer style to go with.

I made a 5 question survey to try and narrow it down.
https://spreadsheets.google.com/viewfor ... VXFnMnc6MQ

I took the survey. Shouldn't maltiness and hoppiness be separated? Hoppiness can also be further broken down to aroma, flavor, and bitterness.

Thank you,

Yes, that would make for a much more evaluative survey but the survey isn't super important it is only to help me narrow down a beer style for Analysis.

So far the porter is winning with IPA a close second.

I've been doing some analysis of porter recipes.

I've attached an image that demonstrates the use of Latent Semantic Analysis with 20 Porter beer recipes. Currently the methodology only produces gibberish at the moment. I still need to add more recipes and adjust them to proportional terms and evaluate the phrases instead of words. But if anything it makes for an interesting graph.