Remoboko

Religion, Morality and Boko in West Africa

Research data, interactive figures and Python code from Remoboko, a study of religiosity and how it affects secular education (boko, in Hausa) in West Africa.

The project focused on the presence, competition and conflict between secularism, Salafism and Pentecostalism on four campuses — Université Abdou Moumouni (Niamey, Niger), University of Ibadan (Nigeria), Université de Lomé (Togo) and Université d'Abomey-Calavi (Benin). It examined how students seeking a degree that would ensure them a better life resort to Salafism and Pentecostalism, and how boko in this context is both appealing and rejected.

Remoboko was a Leibniz Junior Research Group that ran from June 2018 to 2024. It was based at Leibniz-Zentrum Moderner Orient (ZMO) in Berlin and funded through the Leibniz Competition. Project blog: https://remoboko.hypotheses.org/.

Interactive figures

Every figure is a self-contained HTML page served from GitHub Pages, listed at https://fmadore.github.io/Remoboko/. Each page reads its data from the JSON files in this repository at load time, so a data edit is live as soon as it is pushed. The pages are made to be embedded: drop any of the URLs below into an <iframe> and the figure fills the frame.

Maps and timeline for the book:

• Points of interest around Abomey-Calavi and Lomé — 33 churches, mosques, schools, universities and landmarks, searchable and filterable by country (also at points_of_interest.html)
• Universities of Benin and Togo
• Religion, education and politics in Togo and Benin, 1960–2010 — 37 events, filterable by theme

Final report:

• Collaborators by country · by affiliation, on a map · by gender
• Publications and activities by type and language · over time

Every chart offers a table view, CSV download, and PNG/SVG export of the current selection. Timeline themes, activity periods/hidden series, and treemap drilldowns are reflected in the URL for sharing. Maps offer a keyboard-accessible data table, search/selection controls, and a readable fallback when WebGL, the map library or the basemap service is unavailable. Maps use OpenFreeMap vector tiles rendered with MapLibre GL, which need no API key.

Contents

assets/

The interactive figures share small ES modules:

Module	Responsibility
remoboko.css	Design tokens, figure layouts, keyboard focus, tables and map styles
tokens.js	Country/categorical colours, basemaps and source attribution
data.js	Strict calendar dates, normalization, counts and complete time periods
remoboko.js	Data loading, controls, tooltips, tables and responsive figure helpers
chart.js	Chart lifecycle: redraws on resize, font load and export, kept keyboard focus, tooltips, bar shapes
export.js	Self-contained chart SVG/PNG exports, current-view metadata and CSV
maps.js	Map data identities, validation, accessible fallback and shared map lifecycle
d3.js	One pinned D3 import used by every chart

Charts use D3 and maps use MapLibre GL, loaded from a CDN as ES modules; there is no application build step. npm dependencies provide the same releases locally for tests, so regression tests do not need live map tiles or CDN responses.

Book_DeGruyter/

Figures and data for the monograph:

Frédérick Madore, Religious Activism on Campuses in Togo and Benin: Christian and Muslim Students Navigating Authoritarianism and Laïcité, 1970–2023. ZMO-Studien 48. Berlin: De Gruyter, 2025. https://doi.org/10.1515/9783111428895

Path	What it is	View
Maps/UAC_UL_locations_map.html + locations-map.js	33 points of interest around Abomey-Calavi and Lomé, from locations.json	Open ↗
Maps/points_of_interest.html	Same figure at its original URL	Open ↗
Maps/universities_map.html + universities-map.js	The four Beninese and Togolese universities, from universities.json, with logos as markers	Open ↗
Timeline/index.html + script.js	Interactive timeline of the 37 events in data.json	Open ↗
Timeline/timeline.py	Print versions: Religion_Timeline and Education_Politics_Timeline (PNG + SVG)	—

Final report/

Figures and data for the project's final report:

Path	What it is	View
collaborators_by_country.html + .js	93 collaborators across 24 countries, names on hover	Open ↗
collaborators_map.html + .js	The same collaborators located at 56 institutions	Open ↗
collaborators_gender.html + .js	The same collaborators by gender, as a proportion bar	Open ↗
treemap_chart.html + .js	181 publications and activities by type, language and year	Open ↗
activities_type_over_time.html + .js	The same outputs stacked by type, per quarter or per year	Open ↗
collaborators_gender.py	collaborators_gender.png (+ _white variant), the print proportion bar	—
word_clouds.py	WordClouds/english_wordcloud.png and french_wordcloud.png	—

viz_common.py

Dependency-free Python helpers: strict JSON/calendar-date loading, missing-gender normalization, and print palettes matching browser country and gender identities. The book timelines retain their monochrome publication styling.

Data

File	Records
Book_DeGruyter/Maps/locations.json	33 GeoJSON points — name, country, type
Book_DeGruyter/Maps/universities.json	4 GeoJSON points — name, country, city, logo file
Book_DeGruyter/Timeline/data.json	37 events — date, country, category (Religion / Education / Politics), plus optional label-layout hints for the print version
Final report/Data/Collaborators_data.json	93 collaborators — name, gender, affiliation, country, coordinates
Final report/Data/Publications_and_activities_data.json	181 records — title, type, author(s), date, language, abstract

Getting started

Serve the repository root with any static server, for example the bundled one:

python scripts/serve.py --port 8765

Then visit http://localhost:8765/. Opening HTML directly from disk does not work because pages fetch their JSON data.

Tests and validation

Use Node.js 22.22.2 or later. The browser suite covers desktop/phone Chromium, plus a focused Firefox/WebKit compatibility suite. It checks interactions, filtered exports, keyboard access, failure fallbacks, and iframe layouts.

npm ci
npm run check
npx playwright install chromium firefox webkit
npm test

Browser tests serve the pinned npm releases locally and replace external basemaps with an empty style. This checks our rendering and behavior independently of third-party availability; it does not certify that a remote tile service is online. Playwright starts scripts/serve.py, the standard-library server with fixed JavaScript module MIME types (Windows otherwise serves .mjs as text). Reports and failure traces are retained in CI.

Research-data validation uses only Python's standard library:

python scripts/validate_data.py
python -m unittest discover -s tests/python -v

The six schemas in schemas/ cover research JSON and the social-preview boundaries. Validation checks required fields, controlled categories, real calendar dates, numeric coordinate bounds, duplicate identities, polygon closure and local logo references. The bundled evaluator supports the schema keywords used here and rejects unsupported keywords. A full JSON Schema implementation can also consume the individual schema files. Coordinate bounds and valid dates do not verify historical accuracy.

Print figures

Use Python 3.12 or later and a virtual environment. The core constraints capture the tested Python 3.12 Linux rendering environment that CI uses. They constrain core figure dependencies only, not the optional NLP pipeline.

python -m venv .venv
source .venv/bin/activate       # macOS / Linux
# Windows: .venv\Scripts\activate
python -m pip install -r requirements.txt -c constraints-core.txt

python Book_DeGruyter/Timeline/timeline.py --output-dir build/figures
python "Final report/collaborators_gender.py" --output-dir build/figures
python .github/assets/social_preview.py --output-dir build/figures

The generators default to headless output and never render on import. Without --output-dir, they save beside their original script (preserving existing paths). Timeline and gender scripts accept --show for an interactive window and --data for an alternate dataset. Book fonts are optional licensed local files; otherwise the timeline uses DejaVu Sans.

For linting and the seeded word-cloud rendering test:

python -m pip install -r requirements-dev.txt -c constraints-core.txt
ruff check .
python -m unittest discover -s tests/python -v

The small WordCloud renderer is included in development dependencies so CI can test deterministic rendering from a count fixture without downloading language models. Core CI writes figures outside the checkout and retains them as artifacts.

Optional word clouds

NLP dependencies are separate from core plotting. Resource installation is an explicit network operation; ordinary generation uses local resources only.

python -m pip install -r requirements-nlp.txt
python "Final report/word_clouds.py" --download-resources
python "Final report/word_clouds.py" --output-dir build/wordclouds
python "Final report/word_clouds.py" --weighting document --output-dir build/wordclouds

Setup installs NLTK punkt_tab, stopwords, and wordnet, plus the exact fr_core_news_lg 3.8.0 model compatible with spaCy 3.8. The French pipeline retains its tokenizer, morphology and lemmatization components and excludes unused parsing/NER. Texts are batched. A normal run fails with setup instructions if resources are unavailable; it never downloads them implicitly.

Each run writes term/count CSVs and wordcloud_methods_token.json or wordcloud_methods_document.json, alongside the PNGs. The manifest records source and resource hashes, coverage, Python/package/model versions, pipeline components, font hash, seed, and weighting. Document-frequency filenames include _document to coexist with token-frequency outputs. --languages English processes only English; --frequencies-only omits image rendering; --seed and --font make rendering choices explicit.

The optional NLP stack has version ranges and a pinned French model, not a fully tested transitive lock. The manifest records the actual environment. Reproducing a published cloud requires preserving its source file, manifest, dependencies, resources and font; a random seed alone does not guarantee identical output across environments.

Research methods and provenance

The JSON files remain the primary records. Country means the recorded affiliation's country in collaborator figures, not nationality. Missing gender remains visible as Unknown and contributes to the denominator; no gender is inferred. Counts describe the recorded project corpus, not all possible collaborators or activities.

Currently 72 of 181 output records have no abstract. English word-cloud input covers 74 of 114 English records; French covers 34 of 65. German has one nonempty abstract across two records and is not processed by the English/French pipeline. The Abstract field includes both short abstracts and long blog text, so token frequency gives longer records more influence. Document frequency counts each term once per included record and offers a different, explicit weighting.

The pipeline lowercases text, applies language-specific stopwords and lemmatization, and publishes exact counts. English retains the existing WordNet noun-lemma method; French uses the pinned spaCy model. WordCloud receives counts directly, without a second tokenizer, plural normalization or stopword pass. The clouds are exploratory illustrations; the CSVs expose the quantities behind them.

Timeline dates are displayed as recorded. Some dates fall on 1 January, but the source data does not distinguish exact dates from year-only estimates. No precision or missing citations have been invented. The book provides the overarching scholarly source; item-level source identifiers, temporal precision and provenance can be added through documented future curation. Technical validation cannot replace that historical review.

How to cite

If you use this repository — its code, its data or the figures it generates — please cite it. GitHub renders a ready-made citation from CITATION.cff via the Cite this repository button in the sidebar.

For the research itself, cite the book: https://doi.org/10.1515/9783111428895.

License

• Code (Python scripts, viz_common.py, the JavaScript and CSS in assets/ and next to each figure, HTML scaffolding, tests) — MIT.
• Research data and generated figures (the JSON files, and the PNG / SVG outputs) — CC BY 4.0.
• The university logos in Book_DeGruyter/Maps/ and the national flag icons in Book_DeGruyter/Timeline/ are third-party material and are covered by neither licence. The map pin icons in assets/remoboko.js are from Font Awesome Free (CC BY 4.0).

See NOTICE.md for the file-by-file breakdown.

Acknowledgements

Funded through the Leibniz Competition and hosted by Leibniz-Zentrum Moderner Orient (ZMO), Berlin. Country boundaries in .github/assets/ come from Natural Earth (public domain). Basemaps by OpenFreeMap, © OpenMapTiles, data from OpenStreetMap.