Notebooks for the NEH TAPI Workshop, "How to Do Things with Topic Models."
The code for these workshops are found in a collection of Jupyter Notebooks hosted on the Constellete Binder hub. To access them, click on the button below:
Corpus data files are CSV files in a specific format — machine learning corpus format. Files in this format are essentially files with three columns: (1) a document identity, called doc_key, (2) a document label, called doc_label, and (3) the document content, called doc_content. Each row contains a complete "document," defined as an analytically useful unit of discourse, such as a paragraph or chapter.
A collection of them have been prepared for this workshop. Download them and then upload them to the ./corpora folder in your Binder repository. Sorry that the process is not more direct!
Corpus data may be downloaded from the following shared Dropbox link:
Additionally, these files may be downloaded individually:
- Wine Reviews — A collection of terse wine reviews.
- JSTOR Hyperparameter — Abstracts from a JSTOR search for "hyperparameter."
- Tamilnet — A sample of news stories from the website Tamilnet.
- Anphoblach — A sample of news stories from the website Anphoblacht.
Each link goes to a Dropbox item that has a download link. Download the file to your desktop and then upload to the appropriate directory.
The notebooks in this workshop will generate a digital analytical edition from a given source corpus file. The results of the various analytical processes will be put in the ./db directory. A demonstration edition is provided for one of the notebooks. To get the demo data, download then upload the files with the prefix jstor_hyperparameter_demo into to your ./db directory.