vsm.model.LdaCgsMulti

class vsm.model.LdaCgsMulti(corpus=None, context_type=None, K=20, V=0, alpha=[], beta=[])

An implementation of LDA using collapsed Gibbs sampling with multi-processing.

Methods

__init__([corpus, context_type, K, V, ...])	Initialize LdaCgsMulti.
load(filename)	A static method for loading a saved LdaCgsMulti model.
save(filename)	Saves the model in an .npz file.
train([n_iterations, verbose, n_proc, seeds])	Takes an optional argument, n_iterations and updates the model n_iterations times.

__init__(corpus=None, context_type=None, K=20, V=0, alpha=[], beta=[])

Initialize LdaCgsMulti.

Parameters:

• corpus (Corpus) – Source of observed data.
• context_type (string, optional) – Name of tokenization stored in corpus whose tokens will be treated as documents.
• K (int, optional) – Number of topics. Default is 20.
• beta (list, optional) – Topic priors. Default is 0.01 for all topics.
• alpha (list, optional) – Context priors. Default is a flat prior of 0.01 for all contexts.

static load(filename)

A static method for loading a saved LdaCgsMulti model.

Parameters:filename (string) – Name of a saved model to be loaded. Returns:m : LdaCgsMulti object See Also:numpy.load

save(filename)

Saves the model in an .npz file.

Parameters:filename (string) – Name of a saved model to be loaded. See Also:numpy.savez

train(n_iterations=500, verbose=True, n_proc=2, seeds=None)

Takes an optional argument, n_iterations and updates the model n_iterations times.

Parameters:

• n_iterations (int, optional) – Number of iterations. Default is 500.
• verbose (boolean, optional) – If True, current number of iterations are printed out to notify the user. Default is True.
• n_proc (int, optional) – Number of processors used for training. Default is 2.
• seeds (list of integers, optional) – List of random seeds, one for each thread. The length of the list should be same as n_proc. Default is None.