def load_matrix(matrix_name):
matrices = bk._get_matrices()
filename = matrices.get(matrix_name, 'filename')
model_type = matrices.get(matrix_name, 'model_type')
if model_type == 'lsa':
return lsa.LsaModel.load_matrix(filename)
return model.Model.load_matrix(filename)
def viewer(matrix_name):
matrices = bk._get_matrices()
corpus_name = matrices.get(matrix_name, 'vsm_corpus')
corpus = corpus_io.load_corpus(corpus_name)
matrix = model_io.load_matrix(matrix_name)
model_type = matrices.get(matrix_name, 'model_type')
if model_type == 'tf':
return tfviewer.TfViewer(corpus=corpus, matrix=matrix)
elif model_type == 'tfidf':
return tfidfviewer.TfIdfViewer(corpus=corpus, matrix=matrix)
elif model_type == 'lsa':
return lsaviewer.LsaViewer(corpus=corpus, svd_matrices=matrix)
elif model_type == 'beagle-environment':
return envv.BeagleEnvironmentViewer(corpus=corpus, matrix=matrix)
elif model_type == 'beagle-context':
return ctxv.BeagleContextViewer(corpus=corpus, matrix=matrix)
elif model_type == 'beagle-order':
return ordv.BeagleOrderViewer(corpus=corpus, matrix=matrix)
elif model_type == 'beagle-composite':
return comv.BeagleCompositeViewer(corpus=corpus, matrix=matrix)
def train_model(matrix_name):
matrices = bk._get_matrices()
corpus_name = matrices.get(matrix_name, 'vsm_corpus')
corpus = corpus_io.load_corpus(corpus_name)
filename = matrices.get(matrix_name, 'filename')
model_type = matrices.get(matrix_name, 'model_type')
if model_type == 'tf':
m = tf.TfModel()
tok_name = matrices.get(matrix_name, 'tok_name')
m.train(corpus, tok_name)
elif model_type == 'tfidf':
m = tfidf.TfIdfModel()
tf_file = matrices.get(matrix_name, 'tf_matrix')
tf_matrix = model.Model.load_matrix(tf_file)
m.train(tf_matrix=tf_matrix)
elif model_type == 'lsa':
m = lsa.LsaModel()
td_file = matrices.get(matrix_name, 'td_matrix')
td_matrix = model.Model.load_matrix(td_file)
k_factors = matrices.getint(matrix_name, 'factors')
m.train(td_matrix=td_matrix, k_factors=k_factors)
elif model_type == 'beagle-environment':
m = beagleenvironment.BeagleEnvironment()
n_columns = matrices.getint(matrix_name, 'n_columns')
m.train(corpus, n_columns=n_columns)
elif model_type == 'beagle-context':
m = beaglecontext.BeagleContext()
env_file = matrices.get(matrix_name, 'env_matrix')
env_matrix = model.Model.load_matrix(env_file)
m.train(corpus, env_matrix=env_matrix)
elif model_type == 'beagle-order':
m = beagleorder.BeagleOrder()
env_file = matrices.get(matrix_name, 'env_matrix')
env_matrix = model.Model.load_matrix(env_file)
lmda = matrices.getint(matrix_name, 'lambda')
m.train(corpus, env_matrix=env_matrix, lmda=lmda)
elif model_type == 'beagle-composite':
m = beaglecomposite.BeagleComposite()
ctx_file = matrices.get(matrix_name, 'ctx_matrix')
ctx_matrix = model.Model.load_matrix(ctx_file)
ord_file = matrices.get(matrix_name, 'ord_matrix')
ord_matrix = model.Model.load_matrix(ord_file)
m.train(corpus, ctx_matrix=ctx_matrix, ord_matrix=ord_matrix)
m.save_matrix(filename)
def matrix_names():
matrices = bk._get_matrices()
return matrices.sections()