Sagaustus/adh-dh-notebooks

Computational Text Analysis for African Digital Humanities — 112 runnable notebooks organised by method, data served from CourseHub # Computational Text Analysis for African Digital Humanities **112 runnable notebooks**, organised by method. Every one executes top to bottom — enforced by the build, not assumed. Companion to the course at coursehub.africandigitalhumanities.ca, where the same notebooks run in the browser with no setup. ## The data loads itself Open any notebook in Colab and run the first cell. It downloads what it needs from `coursehub.africandigitalhumanities.ca/ide/pack-data/text/` — **no uploading, no API key, no account.** | Corpus | What it is | Licence | |---|---|---| | `masakhanews.csv` | News in 16 African languages with topic labels | AFL-3.0 | | `masakhaner.csv` | MasakhaNER 2.0 — 147,053 sentences, 19 languages, official splits | CC BY 4.0 | | `colonial_texts.csv` | Passages from pre-1929 writing about Africa | Public domain | | `african_legal.csv` | ECOWAS, ANC and TRC documents | Public / open | | `labse_headlines.npy` | Precomputed LaBSE vectors, so neural lessons work with no GPU | derived | ## Where the GPU comes from - **In Colab** — your own free GPU (`Runtime > Change runtime type > T4`). - **In CourseHub** — a shared L40S, via an endpoint whose credential stays on the server. - **Anywhere else** — precomputed vectors shipped with the course. No API key appears anywhere in this repository, by design: a notebook cell is readable by whoever runs it. ## Contents ### 00-foundations Where the field came from, and how to turn a humanities question into a countable one. | Notebook | Method | Open | |---|---|---| | Text as Data | Turning a corpus into a table: document, token, type | | | Practice: Your First Hour With an Unfamiliar Corpus | A reproducible first-look protocol before any analysis | | | What a Corpus Can and Cannot Answer | Matching a literary or linguistic question to a countable method | | | Whose Africa Is in the Public Domain? | Corpus construction as an argument; the absence audit for text | | | Practice: Match Your Question to a M …