wordcounts_instances: Create MALLET instances from a word-counts data frame

In agoldst/dfrtopics: Tools for exploring topic models of text

Create MALLET instances from a word-counts data frame

Description

Given a data frame representing documents as feature counts, create a MALLET InstanceList object which can then be passed on to train_model or saved to disk for later use with write_instances. This function is a small convenience wrapper for make_instances that ensures no further stopword removal, tokenization, or casefolding is done.

Usage

wordcounts_instances(
  counts,
  shuffle = FALSE,
  sep = " ",
  token_regex = "\\S+",
  preserve_case = TRUE
)

Arguments

counts	data frame with id, word, weight columns
shuffle	randomize word order before passing on to MALLET? (See wordcounts_texts
sep	separator to use between words
token_regex	regular expression matching a token. Ordinarily, this should correspond to sep (hence the default, whitespace tokenization), since no further tokenization should be done.
preserve_case	if FALSE, all words are lowercased by MALLET

Details

If your tokens themselves contain whitespace, change the sep parameter and adjust the token_regex accordingly.

Value

an rJava reference to a MALLET InstanceList

See Also

make_instances which this wraps, train_model, write_instances

agoldst/dfrtopics documentation built on July 15, 2022, 4:13 p.m.