In kgjerde/corporaexplorer: A 'Shiny' App for Exploration of Text Collections

knitr::opts_chunk$set(echo = TRUE)

corporaexplorer can be used to explore not only chronological text collections with document date as main organising principle, but any collection of texts. The example used here: Jane Austen's six novels, accessed through the janeaustenr package.

To run the Jane Austen demo app, run the following in the R console:

library(corporaexplorer)
run_janeausten_app()

This app is created with the code below.

Loading packages

library(janeaustenr)
library(dplyr)
library(stringr)
library(tidyr)
library(corporaexplorer)

Loading the Jane Austen books from the janeaustenr package

books <- austen_books()

Pre-processing the text (a bit quick and dirty)

# Regular expression to identify where new chapters begin
chapter_regex <- "((Chapter|CHAPTER|VOLUME) (\\d+|[IXVL])+)"

# Pre-processing
books <- books %>%
    dplyr::group_by(book) %>%
    # Each book into one long string:
    dplyr::summarise(Text = paste(text, collapse = " "), .groups = "drop_last") %>%
    # Insert placeholder at beginning of each chapter
    mutate(Text = str_replace_all(Text, chapter_regex, "NEW_CHAPTER\\1")) %>%
    # Replace double space with two newlines (to restore structure of the text):
    mutate(Text = stringr::str_replace_all(Text, "  ", "\n\n")) %>%
    # Split each book into a character vector (one element is one chapter):
    mutate(Text = stringi::stri_split_regex(Text, "NEW_CHAPTER")) %>%
    # Remove first element (book title), so the books start with Chapter 1
    mutate(Text = lapply(Text, function(x) x[-1]))

# The result is a data frame with one row for each book.
    # The "Text" column is a list of character vectors
    # The "book" column is the name of the book

# From one row per book to one row per chapter
books <- tidyr::unnest(books, Text)

Creating corporaexplorerobject of the Jane Austen books

When we first have a data frame with text and metadata (in this case just book title), creating a "corporaexplorerobject" for exploration is very simple:

# As this is a corpus which is not organised by date,
  # we set `date_based_corpus` to `FALSE`.
# Because we want to organise our exploration around Jane Austen's books,
  # we pass `"book"` to the `grouping_variable` argument.
jane_austen <- prepare_data(
    dataset = books,
    date_based_corpus = FALSE,
    grouping_variable = "book"
)

Run corpus explorer

explore(jane_austen)

Example: "death" in Jane Austen's books:

knitr::include_graphics("../man/figures/austen_death.png")

kgjerde/corporaexplorer documentation built on July 3, 2023, 7:02 p.m.