dictionary-model 0.2.4

Model for tracking context of utterance and predicting future characters.

To use, first create CharPredictor object:

>>> predictor = CharPredictor()

This may take a while as model is being downloaded and loaded.

Then, to track utterance context, you can add letter index to context:

>>> letter_index = 1    # 1 -> a,   letters should be indexed in order: ' abcdefghijklmnopqrstuvwxyz' (0 -> space)
>>> predictor.add_to_context(letter_index)

or you can add letter as string of length 1 (make sure it is one of AsciiEncoder.AVAILABLE_CHARS):

>>> letter = 'a'
>>> predictor.add_to_context(letter)

or you can add probability distribution for all AsciiEncoder.AVAILABLE_CHARS letters:

>>> import numpy as np
>>> import AsciiEncoder as AE
>>>
>>> num_chars = len(AE.AVAILABLE_CHARS)
>>> letter_distr = np.random.random((1, num_chars)) # random proba distribution
>>> predictor.add_to_context(letter_distr)

And finally - you can predict probabilities of each letter coming next after text stored in context. (Letters are indexed in order shown below):

>>> predictor.transform()

Letters order:

' abcdefghijklmnopqrstuvwxyz' # space character comes at index 0, then alphabetical order for indices from 1 to 26