They have an article about K-star rating systems [0] which uses Bayesian approximation [1] [2] (something I know little to nothing about, I'm just regurgitating the article).
There's a whole section on their website that has different statistics for programmers, including rating systems [3].
The formula still works for scales of 5 or 10, you just have to divide by the max rating first and then multiply by it again at the end.
For example a 3/5 stars turns into 0.6 positive and 0.4 negative observation. Following the formula from there will give a lower bound estimation between 0 and 1, so then you just multiple by 5 again to get it between 0 and 5.
* Included a graph of the resulting ordering of the two dimensional plane and some examples
* Included consideration of 5- or 10-star scales.