README.Rmd

---
title: "Wordcloud-Script"
output: github_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

R script to create wordcloud from text document, using the `wordcloud2` package. 

## Example
Here is an example of how to create a wordcloud. I use my PhD thesis as the example text. My PhD thesis can be found at:
<https://www.researchgate.net/publication/319876899_Corporate_networks_of_international_investment_and_trade>

First, load the relevant libraries:
```{r LIB,eval=FALSE}
library(tm)
library(magrittr)
#library(devtools)
#devtools::install_github("lchiffon/wordcloud2")
library(wordcloud2)
```

Then I save the document as a .txt file and then read into R:

```{r Read,eval=FALSE}
text1<-"C:\\Users\\example.txt"
text2 <-readLines(text1)
DOCtxt<-Corpus(VectorSource(text2))
```

I then clean the document, removing whitespace, common words and numbers. I convert all words to lowercase. And finally, I remove a list of words - these are common words for this particular document:

```{r remove,eval=FALSE}
DOCtxt_data<-tm_map(DOCtxt,stripWhitespace)%>%tm_map(.,tolower)%>%
  tm_map(.,removeNumbers)%>%tm_map(.,removePunctuation)%>%
  tm_map(.,removeWords,stopwords("english"))%>%
  tm_map(.,removeWords,c("and","the","our","that","table","figure",
                         "for","are","also","more","has",
                         "must","have","should","this",
                         "with","furthermore","PhD",
                         "additionally","in","number",
                         "which","patterns","use","one",
                         "approach","whether","capture",
                         "effects","specifically",
                         "instance","yet","two",
                         "rather","approaches","along",
                         "however","extent","provides",
                         "applied","characterised",
                         "amongst","need","often","order",
                         "table","based","key","examining",
                         "captures","important","positive","using",
                         "negative","better","perspective","journal",
                         "case","therefore","can","work","et","al",
                         "used","role","set","studies","measures",
                         "paper","provide","figures","tables"))
```

From this - we create an object providing words and frequency:
```{r DTM,eval=FALSE}
dtm<- TermDocumentMatrix(DOCtxt_data)%>%as.matrix()
v <- sort(rowSums(dtm),decreasing=TRUE)
d <- data.frame(word = names(v),freq=v)
```

## Wordcloud Plot
Now I have the clean object, I can create the plots. 

### Wordcloud plot:
```{r WC_command,eval=FALSE}
wordcloud2(d)
```
  
```{r WC, echo=FALSE}
library(NLP)
library(tm)
library(magrittr)
speech <-"C:\\Users\\matt_\\Documents\\Wordcloud\\CORPORATE NETWORKS OF INTERNATIONAL_WORDCLOUD.txt"
THESIStxt <-readLines(speech)
THESIS<-Corpus(VectorSource(THESIStxt))
THESIS_data2<-tm_map(THESIS,stripWhitespace)%>%tm_map(.,tolower)%>%
  tm_map(.,removeNumbers)%>%tm_map(.,removePunctuation)%>%
  tm_map(.,removeWords,stopwords("english"))%>%
  tm_map(.,removeWords,c("and","the","our","that","table","figure",
                         "for","are","also","more","has",
                         "must","have","should","this",
                         "with","furthermore","PhD",
                         "additionally","in","number",
                         "which","patterns","use","one",
                         "approach","whether","capture",
                         "effects","specifically",
                         "instance","yet","two",
                         "rather","approaches","along",
                         "however","extent","provides",
                         "applied","characterised",
                         "amongst","need","often","order",
                         "table","based","key","examining",
                         "captures","important","positive","using",
                         "negative","better","perspective","journal",
                         "case","therefore","can","work","et","al",
                         "used","role","set","studies","measures",
                         "paper","provide","figures","tables"))

dtm<- TermDocumentMatrix(THESIS_data2)%>%as.matrix()
v <- sort(rowSums(dtm),decreasing=TRUE)
d <- data.frame(word = names(v),freq=v)
library(wordcloud2)

PLOTcloud<-wordcloud2(d)
library(htmlwidgets)
library(webshot)

saveWidget(PLOTcloud,"wordcloud_example.html",selfcontained = F)
webshot(url = "wordcloud_example.html", file = "wordcloud_example.png",delay=30) 
         
```
  

### Letterplot:
```{r LC_command,eval=FALSE}
letterCloud(d, word = "EXAMPLE", color='random-light' , backgroundColor="black")
```
  
  
```{r WL, echo=FALSE}
library(NLP)
library(tm)
library(magrittr)
speech <-"C:\\Users\\matt_\\Documents\\Wordcloud\\CORPORATE NETWORKS OF INTERNATIONAL_WORDCLOUD.txt"
THESIStxt <-readLines(speech)
THESIS<-Corpus(VectorSource(THESIStxt))
THESIS_data2<-tm_map(THESIS,stripWhitespace)%>%tm_map(.,tolower)%>%
  tm_map(.,removeNumbers)%>%tm_map(.,removePunctuation)%>%
  tm_map(.,removeWords,stopwords("english"))%>%
  tm_map(.,removeWords,c("and","the","our","that","table","figure",
                         "for","are","also","more","has",
                         "must","have","should","this",
                         "with","furthermore","PhD",
                         "additionally","in","number",
                         "which","patterns","use","one",
                         "approach","whether","capture",
                         "effects","specifically",
                         "instance","yet","two",
                         "rather","approaches","along",
                         "however","extent","provides",
                         "applied","characterised",
                         "amongst","need","often","order",
                         "table","based","key","examining",
                         "captures","important","positive","using",
                         "negative","better","perspective","journal",
                         "case","therefore","can","work","et","al",
                         "used","role","set","studies","measures",
                         "paper","provide","figures","tables"))

dtm<- TermDocumentMatrix(THESIS_data2)%>%as.matrix()
v <- sort(rowSums(dtm),decreasing=TRUE)
d <- data.frame(word = names(v),freq=v)
library(wordcloud2)

PLOTletter<-letterCloud(d, word = "EXAMPLE", color='random-light' , backgroundColor="black")
library(htmlwidgets)
library(webshot)
saveWidget(PLOTletter,"letter_example.html",selfcontained = F)
webshot(url = "letter_example.html", file = "letter_example.png",delay=120)
```
  
## Writing Plots to files  
These plots output to html, where you can then save as an image in your web browser. Otherwise, you can save these as images directly in R, using `htmlwidget` and `webshot` packages.

```{r SAVE_image,eval=FALSE}
#install webshot
library(webshot)
webshot::install_phantomjs()

## Make the wordcloud
CLOUD<-wordcloud2(d)

## save it in html
library("htmlwidgets")
saveWidget(CLOUD,"CLOUD.html",selfcontained = F)

## Save as image (png, jpeg or pdf)
###It is important to specify a delay - as this gives time for the wordcloud to load fully in html

webshot("CLOUD.html","CLOUD_1.png", delay =15)

webshot("CLOUD.html","CLOUD_1.pdf", delay =15)


```