Categories
Beeributes
Website Comments by FLASHPRO
Now that I'm actually trying to use the "beeributes" feature on BP 3.0, I'm seeing how annoying it is to use. The purpose of it is to track attributes of every beer so that it provides more refinement that just a style. What that means is with enough data, it's possible to match beers up with other similar beers, for recommendations. Eventually, I wanted to use Machine Learning to predict what a person will like. The concept is pretty simple with the technology available with services like Amazon Web Services. You take a ton of known data and run it through some machine learning algorithms that are pretty CPU intensive, and then you have a model that can be applied to individual users or beers in just a fraction of time. Now I'm back to figuring out how to make it happen? I'll continue in a new reply...
3 years ago
Right now the Beeributes presents the user with a long list of pre-defined "tags" that are the beeributes. These tags came from beer flavor wheels that have been around for a long time. They're a more scientific approach to the actual flavors you can detect in a beer, but not really useful or easy to process.
One basic question is "does it make sense to let users write in their own tag?" The advantage is you're not stuck with the same old list of things that may be missing some valuable attributes. The problem is it's likely there would be a lot of crap added if manual entry was allowed, like people saying it tastes like "dog crap" or "unicorns". Or maybe it's ok to just let that stuff slide?
And then there's the usability of it. Having to read through a long list of tags like that is a turn off and confusing, especially if you don't know exactly what word is available. So I think a major improvement would be to allow typing in words and letting it autocomplete with the available attributes. I think it would also be much nicer to have a list of tags that are grouped in some sort of logical way, but I have a feeling that's easier said that done.