DTM: Document Term Matricizer

In myeomans/DTMtools: Adjustable Feature Counting for Text Analysis

Description

Turns text into data.

Usage

DTM(
  texts,
  sparse = 0.99,
  wstem = "all",
  ngrams = 1,
  language = "english",
  vocabmatch = NULL,
  stop.words = TRUE,
  punct = FALSE,
  POS = FALSE,
  dependency = FALSE,
  tag.sub = 0,
  overlap = 0.8,
  group.conc = NULL,
  group.conc.cutoff = 0.8,
  TPformat = FALSE,
  verbose = FALSE,
  mc.cores = 1
)

Arguments

texts	a character vector of texts.
sparse	maximum feature sparsity for inclusion (1 = include all features)
wstem	character what words should be stemmed?
ngrams	numeric vector of ngram sizes (max = 1:3)
language	character what language are you parsing?
vocabmatch	matrix used to create a new matrix with features that are identical to a previous one
stop.words	logical should stop words be included? default is TRUE
punct	logical should exclamation points and question marks be included as features?
POS	logical should features have part of speech tags appended? default is FALSE
dependency	logical should features have dependency relations appended? default is FALSE
tag.sub	numeric what fraction of features should be replaced by POS tags? default is 0 (no features), fractions not supported yet.
overlap	numeric How dissimilar (in cossine distance) must an ngram be from all (n-1)grams to be added to feature set?
group.conc	character group IDs for removing group-specific words
group.conc.cutoff	numeric threshold for group-specificity of words, as proportion of occurences in the main group.
TPformat	logical - return in stm::textProcessor() format?
verbose	logical - report interim steps during processing

Value

Feature counts, as a matrix (or in stm format)

myeomans/DTMtools documentation built on March 2, 2020, 8:57 p.m.