Clark Pubs: 2022 Journal Analysis

## Loading necessary packages.
library(stringi)
library(SnowballC)
library(tidyverse)
library(tidytext)
library(dplyr)
data("stop_words")
library(ggplot2)
library(wordcloud)

Introduction

This project looks at the word frequency in the 2022 Clark Journal, with the hope that an insight to word frequency would

• consolidate a word-bank for museum writers

• provide keywords of institutional identity

• reflect broader lexicon trend (presence AND absence of certain words) in the industry

I omitted five types of article in this analysis: 8. acquisition checklist, 12. visitor information, 13. donor list, 14. tribute gifts, 15. board of trustees.

I. What does the word frequency look like in each article?

To find out, I wrote four custom functions:

• getDfNontrivialWords: takes a txt file; cleans the text by converting all characters into ASCII encoding (e.g. é to e), deleting all trivial stopwords (e.g. articles, possessives, conjunctions, etc.), removing numbers, and converting every word into its stem (that is, getting the basic conjugate, e.g. exhibitions into exhibit, printmaker into printmak, natural into natur, digitize into digit, etc.); returns a dataframe that contains all the individual, cleaned words in the txt file.

## function: get individual word df
getDfNontrivialWords <- function(txtfile) {
  
## read director's text
text <- readLines(txtfile, encoding = 'UTF-8', skipNul = T)
## convert UTF-8 to ASCII encoding
text <- stri_trans_general(text, "Latin-ASCII")
## into dplyr tibble
df <- tibble(line=1:length(text), text=text)
## separate text into individual words
## delete trivial stopwords
df <- df %>%
  unnest_tokens(word, text) %>%
  anti_join(stop_words)
## remove numbers
df <- df[-grep("\\b\\d+\\b", df$word),]  
## stemming (get basic conjugate)
df <- df %>%
      mutate_at("word", funs(wordStem((.), language="en")))  

return(df)  

}

• getTableWordFreq: takes a txt file; returns a tibble (table) that ranks all the individual, non-numerical, non-trivial words in the txt file in order of frequency.

## function: get word frequency table
getTableWordFreq <- function(txtfile) {
  
df <- getDfNontrivialWords(txtfile)  
dffreq <- df %>%
  count(word) %>%
  arrange(desc(n))

return(dffreq)

}

• getBarplotWordFreq: takes a txt file; returns a barplot that ranks the 20 most frequent individual, non-numerical, non-trivial words in the txt file in order of their frequency.

## function: barplot of word frequency table
getBarplotWordFreq <- function(txtfile){
  
df <- getDfNontrivialWords(txtfile)  
dfPlot <- df %>%
  count(word, sort = TRUE) %>%
  mutate(word = reorder(word, n)) %>%
  slice(1:20) %>%
  ggplot(aes(n, word, fill=word)) +
  geom_bar(stat = "identity") +
  theme(panel.background = element_blank()) +
  theme(legend.position = "none") +
  xlab("Frequency") + ylab("Word Stem") +
  scale_fill_manual(
    values = rep(c("darkgoldenrod2", "lightblue"), 999)
  )
return(dfPlot)

}

• getWordcloud: takes a txt file; returns a wordcloud that ranks all the individual, non-numerical, non-trivial words (display at most 80 words) in the txt file in order of their frequency.

## function: get wordcloud
getWordcloud <- function(txtfile) {
  
palette <- brewer.pal(9, "GnBu")
set.seed(0)
wordcloud <- getTableWordFreq(txtfile) %>% 
  with(wordcloud(word, n, random.order = FALSE, min.freq = 1, max.words = 80, colors=palette))
return(wordcloud)

}

1. Director’s text

2. Dec arts interpretation

3. RAP

4. Library

5. Grounds + WTF

6. Musical Performance

9. Upcoming Exhibitions

10. Advancement

II. What does the word frequency look like for the whole journal?

The top-20 word stems, in order of frequency.

## # A tibble: 20 × 2
##    word         n
##    <chr>    <int>
##  1 clark      109
##  2 art        101
##  3 histori     52
##  4 exhibit     46
##  5 program     43
##  6 collect     42
##  7 museum      42
##  8 artist      40
##  9 project     40
## 10 institut    38
## 11 perform     35
## 12 print       34
## 13 centuri     33
## 14 intern      31
## 15 galleri     30
## 16 music       28
## 17 object      27
## 18 digit       20
## 19 draw        20
## 20 printmak    20

Word cloud, visualizing the frequently-used word stems throughout the entire 2022 journal.