weightdictr's Introduction

weightdictR

The goal of weightdictR is to facialiate the deployment of weighted dictionaries especially ones including complex multiword patterns. The package was build taking inspirations from gruendls package multidictR.

Installation

You can install the development version of weightdictR from GitHub with:

# install.packages("devtools")
devtools::install_github("TomSOWI/weightdictR")

Example

First of all clean and prepare textdata. This is only necessary if the weighted dictionary includes complex multiword-patterns.

library(weightdictR)
library(quanteda)
library(dplyr)

#Takes a long time to execute

toksBT17_20_clean <- corpus_to_compound_tokens(
  corpus = corpBT17_20,
  pattern = popdictR::gruendl_terms,
  include_totals = TRUE)

…this is already done for a sample of 100 documents to run as an example

#Clean and compound textdata sample
weightdictR::toksBT17_20_clean

#This is my dictionary
weightdictR::klotz_clara


#Deploy my weighted dictionary using the main function of this package: run_weightdict

result_klotz <- run_weightdict(
  text = weightdictR::toksBT17_20_clean,
  dict = klotz_clara,
  dict_name = "klotz"
  )

#Individual context information -----------------------------------------------

#Add party 
party <- data.frame(docname = quanteda::docnames(corpBT17_20), party = docvars(corpBT17_20)$party)
result_klotz <- merge(result_klotz, party)

#Add totals
totals <- data.frame(docname = quanteda::docnames(corpBT17_20), nsentences = nsentence(corpBT17_20), ntokens = ntoken(corpBT17_20))
result_klotz <- merge(result_klotz, totals)

#Final result
result <- result_klotz %>%
  group_by(party) %>%
  reframe(
    nsentence = sum(nsentences),
    ntokens = sum(ntokens),
    klotz_clara = sum(weight)/nsentence,
  )
result

#You may continue by comparing the results using z-transformation
result$klotz_clara_z <- scale(result$klotz_clara, center = T)

Recommend Projects