def get_joint_ibm1(e_corpus: Corpus, f_corpus: Corpus):
PL = cat.LengthDistribution()
PM = cat.LengthDistribution()
PZ = cat.ClusterDistribution(1)
PEi = cat.UnigramMixture(1, e_corpus.vocab_size())
PAj = cat.UniformAlignment()
PFj = cat.BrownLexical(e_corpus.vocab_size(), f_corpus.vocab_size())
return JointModel(PL, PM, PZ, PEi, PAj, PFj)
def get_joint_ibm1z(e_corpus: Corpus,
f_corpus: Corpus,
n_clusters=1,
cluster_unigrams=True,
alpha=1.0):
PL = cat.LengthDistribution()
PM = cat.LengthDistribution()
if not cluster_unigrams:
PZ = cat.ClusterDistribution(n_clusters)
else:
PZ = cat.ClusterUnigrams(n_clusters)
PEi = cat.UnigramMixture(n_clusters, e_corpus.vocab_size(), alpha)
PAj = cat.UniformAlignment()
PFj = cat.MixtureOfBrownLexical(n_clusters, e_corpus.vocab_size(),
f_corpus.vocab_size(), alpha)
return JointModel(PL, PM, PZ, PEi, PAj, PFj)
def get_mlp_ibm1(e_corpus: Corpus, f_corpus: Corpus):
PL = cat.LengthDistribution()
PM = cat.LengthDistribution()
PAj = cat.UniformAlignment()
PFj = MLPLexical(e_corpus, f_corpus)
return ConditionalModel(PL, PM, PAj, PFj)
def get_ibm1(e_corpus: Corpus, f_corpus: Corpus):
PL = cat.LengthDistribution()
PM = cat.LengthDistribution()
PAj = cat.UniformAlignment()
PFj = cat.BrownLexical(e_corpus.vocab_size(), f_corpus.vocab_size())
return ConditionalModel(PL, PM, PAj, PFj)