Fix tests for quanteda v4.2.0.
adjust_alpha as an experimental argument to
optimize alpha automatically.update_model to update terms of existing models to
classify documents with unseen words more accurately.std::vector to
arma::mat.perplexity() to compute perplexity scores of fitted
LDA models.alpha and beta to be a vector for
asymmetric Dirichlet priors.uniform to simplify the computation of seed word
weights.levels argument to better handle hierarchical
dictionaries.textmodel_seqlda() is called.auto_iter to textmodel_seededlda() and
textmodel_lda() to stop Gibbs sampling automatically before
max_iter is reached.batch_size to textmodel_seededlda()
and textmodel_lda() to enable the distributed LDA algorithm
for parallel computing.textmodel_seededlda() and
textmodel_lda() for sequential classification.textmodel_seqlda() as as short cut for
textmodel_lda(gamma = 0.5).regularize argument to
divergence() for the regularized topic divergence
measure.data_corpus_moviereviews to the package to reduce
dependency.min_prob and select to
topics() for greater flexibilityweighted, min_size,
select to divergence() for regularized topic
divergence scores.textmodel_seededlda() to set positive integer
values to residual.textmodel_seededlda() that ignores n-grams
when concatenator is not “_“.topics() to return document names.divergence() to optimize the number of topics or
the seed words (#26).model argument to textmodel_lda()
to replace predict().textmodel_seededlda object to save
dictionary and related settings (#18)predict() to identify topics of unseen documents
(#9)dfm_trim() in textmodel_seededlda() via
... (#8)topics() to return factor with NA for empty
documents