In PolMine/UCSSR: Using Corpora in Social Science Research

Loading Packages {.smaller}

library(polmineR)
library(GermaParl)
library(topicmodels)
library(DT)

use("GermaParl")

Szenario: Diktionärsbasierte Frame-Analyse

crime_dict <- c(
  "krimin.*",
  "[dD]ieb.*",
  "Vergewa.*",
  "Übergriff.*",
  ".*[gG]rapsch.*",
  "sexuell.*",
  "[bB]elästig.*",
  ".*[dD]elikt.*",
  ".*[vV]ersto.*",
  "Körperverl.*",
  ".*[sS]traftat",
  "[Kk]lauen",
  "[sS]tehlen",
  "ausrauben",
  "[Dd]rogen"
)

use("MigParl")
x <- context("MIGPARL", query = '".*[Ff]lüchtling.*"', cqp = TRUE)
x2 <- context("MIGPARL", query = '".*[Ff]lüchtling.*"', cqp = TRUE, positivelist = crime_dict, regex = TRUE)

x3 <- context(
  "MIGPARL", query = '".*[Ff]lüchtling.*"', cqp = TRUE,
  left = 15, right = 15,
  positivelist = crime_dict, regex = TRUE
)

kwic(x3, left = 15, right = 15) %>% highlight(yellow = crime_dict, regex = TRUE)

Create s_attribute 'speech' {.smaller}

if (!"speech" %in% s_attributes("GERMAPARL"))
  GermaParl::germaparl_add_s_attribute_speech()

Load topicmodel {.smaller}

lda_dir <- system.file(package = "GermaParl", "extdata", "topicmodels")
files_in_lda_dir <- list.files(lda_dir)

if (!"germaparl_lda_speeches_250.rds" %in% files_in_lda_dir)
  GermaParl::germaparl_download_lda(k = 250)

lda250 <- germaparl_load_topicmodel(k = 250)
lda_250_terms <- topicmodels::terms(lda250, k = 25)
colnames(lda_250_terms) <- gsub("^Topic\\s+(\\d+)$", "\\1", colnames(lda_250_terms))

Inspecting the Topicmodel {.smaller}

DT::datatable(data = t(lda_250_terms)[,1:7])

Encode topicmodel {.smaller}

if (!"topics" %>% s_attributes("GERMAPARL"))
  GermaParl::germaparl_encode_lda_topics(k = 250, n = 5)

XXX {.smaller}

racism <- 65
citizenship_topic <- 133
refugees <- 213
citizenship_speeches <- GermaParl::germaparl_get_speeches_for_topic(n = citizenship_topic)

XXX

lda_terms_250 <- topicmodels::get_terms(lda250, k = 250)
citizenship_terms <- lda_terms_250[, citizenship_topic]

Rede Rudolf Körper

read(citizenship_speeches[[2]]) %>% highlight(yellow = citizenship_terms)

Rede Erwin Marschewski

read(citizenship_speeches[[3]]) %>%
  highlight(yellow = citizenship_terms)

Rede Cem Özdemir

read(citizenship_speeches[[4]]) %>%
  highlight(yellow = citizenship_terms)

Erwin Marschewski II

s_attributes(citizenship_speeches[[3]], "topics")
terms_topic_2 <- lda_terms_250[, 69]
terms_topic_2

read(citizenship_speeches[[3]]) %>%
  highlight(yellow = citizenship_terms, lightgreen = terms_topic_2)

PolMine/UCSSR documentation built on June 13, 2022, 10:23 p.m.