README.md

In dtburk/gensci.stm: Functions to prepare gender and science reports for analysis with STM, and to run that analysis

gensci.stm

This package bundles functions to generate topic models describing gender in science reports using the stm package.

Installation

You can install this package from GitHub with:

# install.packages("remotes")
remotes::install_github("dtburk/gensci.stm")

This will also install a few packages from CRAN which this package uses. However, this package also uses another package that is not on CRAN called texanaaid (TEXt ANAlysis AIDe). Use the same function as above to install texanaaid from GitHub.

remotes::install_github("dtburk/texanaaid")

For more information on texanaaid, check out the GitHub repository.

Usage

This package includes an R Markdown template that lays out all the steps needed to generate a topic model. To use this template, first make sure you have installed the package, then in RStudio, navigate to File > New File > R Markdown…, click on “From Template” on the left side of the dialog box, select “Construct a topic model of gender in science reports”, and click OK.

This will open a new R Markdown file that contains all the code necessary to generate a topic model for gender in science reports, from extracting text from the PDF reports, all the way to producing output to help interpret the model. Before knitting the R Markdown file with RStudio’s “Knit” button, however, you must set the file paths and analysis parameters listed in the section “Set file paths and parameters” of the template.

A good workflow for analysis would be to create a new directory for each new topic model you want to generate with different parameters, then use the included template to create an analysis script in that directory. This will ensure that the code used to generate a topic model is kept tightly bundled with the model output.

Note: The packages gensci.stm and texanaaid hard-code many data-cleaning and analysis decisions that were made iteratively as the analysis of gender in science reports developed. These decisions may need to be revisited if the analysis is extended beyond the original set of documents. Changes can be made by forking the package repositories on GitHub and editing the functions found in the “R” directory of each package.

dtburk/gensci.stm documentation built on Nov. 13, 2019, 12:33 a.m.