version 1.0.7
Features
- Adds
dictionary_meta to assess dictionary
categories.
- Adds dictionary category coverage reporting to
lma_termcat and lma_process.
Bug Fixes
- Correctly applies category selection in
read.dic in
more cases.
Improvements
- Removes option to download embeddings without terms files; works
toward independence between spaces and term map.
- Allows for direct weight discretization from a weighted list
dictionary.
version 1.0.6
Features
- Adds conversion of list-representation to sparse matrix to
lma_dtm.
Improvements
- Adds overwrite options to download functions.
lingmatch more flexibly handles pairwise comparisons
within groups.lma_dtm applies row names where possible.
Bug Fixes
- Corrects handling of multiple groups given
all.levels
in some cases.
- Adjusts
lma_simets pairwise mean for self match.
version 1.0.5
Features
- Adds
report_term_match to assess fuzzy terms in
dictionaries.
Bug Fixes
- Avoids integer overflow when setting up large comparisons.
- Fixes
select.lspace downloaded indicator.
- Catches more invalid regular expressions to avoid crashes from
pattern_search.
version 1.0.4
Improvements
- Avoids checking if long texts are files.
- Improves feedback when a directory has not been specified.
- Changes results of singular pairwise comparisons from
1
to NA.
version 1.0.3
Bug Fixes
- Fixes
lma_simets vector to matrix comparisons in some
cases.
- Fixes
lma_dtm tokens.only when one text
ends up empty.
version 1.0.2
Features
- Adds options to specify term and category names in
read.dic.
- Adds an option to remove unmatched categories or terms in
lma_patcat.
Improvements
- Improves input-handling of dictionaries.
- Enables term exclusions for token-only
lma_dtm.
- Makes output formats more consistent.
- Handles special-character conversion on misencoded text.
- Better handles unrecognized weight and metric names.
- Handles inconsistently named dictionaries.
Bug Fixes
- Fixes
lma_dtm token to dtm conversion when final token
indices entries are empty.
- Appropriately allows
read.dic to read urls.
- Term weights correctly default to count given an empty weight.
- Adds the
pois.x argument to lma_weight to
allow for separately specified augment alphas and d/ppois quantiles or
probabilities.
- Corrects handling of empty texts in some cases.
- Corrects
lma_process routing and argument passing in
some cases.
version 1.0.1
Bug Fixes
- Avoids a compilation issue on older macOS platforms.
- Avoids factor-related issues when the
stringsAsFactors
option is TRUE.