GMTM version 0.3.0
- Add
include_prob to textmodel_gmm() to
reduce the size of textmodel_gmm when it is
FALSE.
- Add
as.textmodel_gmm() and
as.textmodel_kmeans() to upgrade old objects to the current
format.
- Change the structure of objects while providing compatibility with
the earlier versions of the package.
- Change the
omit argument to experimental.
- Remove
filter from terms() in favor of
applying dfm_select() or tokens_select() to
data.
GMTM version 0.2.1
- Improve the consistency of the
textmodel_kmeans and
textmodel_gmm objects.
- Change the method to identify the most likely topic terms for
textmodel_gmm.
- Fix the handling of multi-word expressions in
as.seedwords().
GMTM version 0.2.0
- Add
probability() to extract topic likelihood of
documents.
- Add
seeds to textmodel_kmeans() and
textmodel_gmm() to initialize clustering with user-provided
centers.
- Add
as.seedwords() to create a seed words matrix from
quanteda dictionaries.
GMTM version 0.1.0
- Create
textmodel_kmeans() and
textmodel_gmm() as wrappers around Armadillo’s
kmeans and gmm_diag functions,
respectively.
- Create
terms() and topics() as utility
functions for topic analysis.