Hi,

Since there's no API reference, and the tutorials don't seem to cover much, I'm having a hard time understanding exactly what OpenBayes can do.

This is how I'd like to use a Bayesian classifier:
1. Call train(cls, features), where cls is a tuple of the form ('class name',float), and features is a dictionary of the form {'feature1':1.2, 'feature2':0.3, ..., 'featureN':0.8}. This will work incrementally, so I can add new training data whenever I want without have to "rebuild" the classifier by re-training it on the entire corpus.
2. Call guess(features), where features is the same as above. This returns a list of the most likely classes associated with the given features.

Is this simple usage possible with OB? I've written some simple code to do this, but it currently only handles discrete classes and features. The Orange toolkit has a Bayesian classifier that supports continuous features, but it doesn't support incremental training, so it's essentially useless for any real work.