ttt_fit_topics: ttt_topics

In mpadge/texttimetravel: Tools for analysing temporally structured text collections

Description

Fit sentence-based topic models for all of some of a corpus, with the ability to fit topics around a specified word or phrase.

Usage

ttt_fit_topics(tok, years = NULL, ntopics = 20, topic = NULL,
  filename = NULL, quiet = FALSE)

Arguments

tok	A quanteda 'tokens' object
years	If specified, restrict topic models to specified years only (requires 'tok' to include a 'docvars' of "year" or similar).
ntopics	Number of topics to be fitted. This should be reasonably large, to allow words not strongly associated with any primary topics to be allocated to latter, effectively meaningless topics.
topic	If specified, construct topic models around the specified phrase.
filename	If specified, the topic model is saved to the nominated file and can be re-loaded with 'x <- readRDS(filename)'.
quiet	If 'TRUE', display progress information on screen.

Value

An topicmodels object of class 'LDA'.

Note

This function may take a long time to execute; please be patient.

Examples

library(quanteda)
dat <- data_corpus_inaugural %>% # from quanteda
     corpus_reshape (to = "sentences") # convert documents to sentences
tok <- tokens (dat, remove_numbers = TRUE, remove_punct = TRUE,
               remove_separators = TRUE)
tok <- tokens_remove(tok, stopwords("english"))
x <- ttt_fit_topics (tok, ntopics = 5)
topicmodels::get_terms(x, 20)
x <- ttt_fit_topics (tok, years = 1789:1900, ntopics = 5)

mpadge/texttimetravel documentation built on Nov. 14, 2020, 11:31 a.m.