qualityquantity.Rmd

---
title: "Alle Worte sind gleich?"
subtitle: 'Term-Extraktion mit polmineR'
author: "Andreas Blaette"
date: "Stand: 19. November 2018"
output:
  ioslides_presentation:
    css: css/stylesheet.css
    logo: img/polmine.png
    widescreen: yes
  slidy_presentation: default
editor_options:
  chunk_output_type: console
vignette: >
  %\VignetteIndexEntry{Sag mir welche Nachbarschaft Du hast}
  %\VignetteEncoding{UTF-8}
  %\VignetteEngine{knitr::rmarkdown}
---

## Loading Packages {.smaller}

```{r}
library(polmineR)
library(GermaParl)
library(topicmodels)
library(DT)
```

```{r}
use("GermaParl")
```


## Szenario: Diktionärsbasierte Frame-Analyse

```{r}
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"
)
```

```{r}
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}

```{r}
if (!"speech" %in% s_attributes("GERMAPARL"))
  GermaParl::germaparl_add_s_attribute_speech()
```


## Load topicmodel {.smaller}

```{r, eval = TRUE}
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}

```{r, render = knit_print, echo = FALSE, message = FALSE}
DT::datatable(data = t(lda_250_terms)[,1:7])
```


## Encode topicmodel {.smaller}

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


## XXX {.smaller}

```{r, eval = FALSE}
racism <- 65
citizenship_topic <- 133
refugees <- 213
citizenship_speeches <- GermaParl::germaparl_get_speeches_for_topic(n = citizenship_topic)
```

## XXX

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


## Rede Rudolf Körper 

```{r}
read(citizenship_speeches[[2]]) %>% highlight(yellow = citizenship_terms)
```


## Rede Erwin Marschewski

```{r}
read(citizenship_speeches[[3]]) %>%
  highlight(yellow = citizenship_terms)
```


## Rede Cem Özdemir

```{r}
read(citizenship_speeches[[4]]) %>%
  highlight(yellow = citizenship_terms)
```


## Erwin Marschewski II

```{r}
s_attributes(citizenship_speeches[[3]], "topics")
terms_topic_2 <- lda_terms_250[, 69]
terms_topic_2
```

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