Skip to content

Reading edition

Islam's “Peripheries”: Digital Humanities, Algorithmic Analysis, and AI in West Africa and Central Asia

Aksana Ismailbekova · Frédérick Madore

Digital History in/of Central Asia · FAU Erlangen-Nürnberg · 29 June 2026

24 slides

Contents

Slide 1

Seminar series · Digital History in/of Central Asia · FAU Erlangen-Nürnberg · 29 June 2026

Islam's “Peripheries”

Digital humanities, algorithmic analysis, and AI in West Africa and Central Asia.

Slide 2

Funding · Volkswagen Foundation

An “Open Up” research project

A collaboration funded by the Volkswagen Foundation, under its Open Up — New Research Spaces for the Humanities and Cultural Studies programme.

The programme's wager To “open up” is to take the first step into something new and unknown — the call backs small teams to explore entirely new research spaces, complex topics that need more than one perspective.
Start
Autumn 2026
Duration
18 months
Team
Two researchers · two regions

Slide 3

The conceptual stake

Why “peripheries”?

  • In Islamic studies, sub-Saharan Africa and Central Asia are often treated as marginal
  • Distant from the “real” Islam of the Middle East
  • Yet Muslim delegations from Togo and Dahomey (Benin) visited Soviet Central Asia, 1960s–70s — and vice versa
  • The project challenges the centre–periphery frame itself

Slide 4

In the news · April 2026

Togo ↔ Kyrgyzstan, 2026

A different kind of tie in the present day. In April 2026, Togo's president made the first-ever state visit to Kyrgyzstan — diplomacy and trade now, where the 1960s links were religious.

Faure Gnassingbé, in an embroidered Kyrgyz robe, walks with Kyrgyz officials past traditional musicians during his April 2026 visit to Kyrgyzstan. République Togolaise Faure Gnassingbé à Bichkek First Togolese leader to visit Kyrgyzstan · 29 April 2026 · republicoftogo.com ↗
Presidents Faure Gnassingbé of Togo and Sadyr Japarov of Kyrgyzstan at a formal signing ceremony in Bishkek, 29 April 2026. AKIpress · Bishkek Kyrgyzstan and Togo sign accords Education, trade, health, digitalisation · 29 Apr 2026 · akipress.org ↗

Slide 5

01

The challenge

Vast multilingual archives, studied in silos.

Slide 6

The problem

Vast archives, locked away

~16,000documents
12+languages
Partialcataloguing

Russian Arabic Hausa Ewe Kabyè Tajik French Uzbek Persian Turki German English

Why it stays locked
  • Sheer volume and linguistic diversity defeat traditional analysis
  • No single scholar or team reads every language and script
  • Rich historical insight stays out of researchers' reach

Slide 7

Rarely compared

Studied in silos

West Africa · post-1960s
  • Islamic discourse and public engagement
  • Post-colonial nation-building
  • Western-educated Muslim voices
  • Francophone press and audiovisual sources
Central Asia · Soviet era + Tajik civil war
  • Russian imperial administration records
  • Early Soviet governance of Muslim communities
  • Islamic reformers and local agency
  • Rare Emirate of Bukhara materials
Common comparative ground
  • Secularism
  • Modernity
  • Development
  • Islamic reformism

Slide 8

Both held at ZMO

Two multilingual digital collections

Islam West Africa Collection
(IWAC)
  • 14,500+ items — newspapers, Islamic publications, audiovisual recordings, photographs
  • 9,315 minutes of audio, mostly Hausa and Arabic
  • 6 countries, since the 1960s
Reinhard Eisener
Collection
  • 1,619 documents across 50 archival boxes
  • Bukhara (1917–30), Soviet governance, Tajik civil war (1992–97)
  • 8 languages incl. Turki; multiple scripts
A montage of sample documents from both collections: handwritten manuscripts, an Arabic-script newspaper, and printed francophone Islamic pamphlets, in several languages and scripts.
Samples from both collections.

Slide 9

Open access · islam.zmo.de

The Islam West Africa Collection

Slide 10

A scholar's estate · access on request

The Reinhard Eisener Collection

Slide 11

Fieldwork · Osh, Kyrgyzstan

A possible third collection

Aksana Ismailbekova has been at Osh State University, in southern Kyrgyzstan.

  • The university holds a large collection of Islamic materials
  • Gathered from local people across the region
  • Manuscripts, devotional prints, loose-leaf texts
Islamic manuscripts and prints collected from local communities · Osh State University, Kyrgyzstan. Islamic manuscripts and prints collected from local communities · Osh State University, Kyrgyzstan. Islamic manuscripts and prints collected from local communities · Osh State University, Kyrgyzstan. Islamic manuscripts and prints collected from local communities · Osh State University, Kyrgyzstan.
Islamic manuscripts and prints collected from local communities · Osh State University, Kyrgyzstan.

Slide 12

02

The approach

AI-driven digital humanities, built across two regions.

Slide 13

Research question

How can AI-driven DH methods transform access to and interpretation of these collections — enabling new comparative methods for Islamic discourse across regions?

The project's guiding question

Slide 14

Innovation 1 / 3

From raw documents to structured data

Multimodal LLMs turn scanned pages into structured data

  • Extracts text from scans and audio — Hausa, Arabic, Cyrillic, Old Tatar
  • Now reads handwriting too — cursive letters and manuscripts as well as print
  • Pulls out people, places, dates as structured fields
  • Opens DH methods — maps, networks, analysis at scale
A scanned francophone newspaper article, ‘Organisation du hadj — Les imams se battent’, with photographs and dense multi-column text — the kind of layout that defeats conventional OCR.
A dense multi-column newspaper page — the kind of layout conventional OCR breaks on.

Slide 15

Innovation 1 · in practice

Reading printed Ewé

A 1976 page of the Togolese newspaper Togo Presse, printed in the Ewé language with full diacritics

Togo Presse, 31 July 1976.

Extracted text · Ewé
MIAFE ŊUTIFAFAFIA EYADEMA KPE TA KPLE MƆSLEMTƆWO FE ƉEKAWƆWƆ DƆDZIKPƆHA LA

Mia nɔvi Mɔslem hamea ngɔnɔlawo se nya xɔasiwo tso Togo yeyea fofo nu

«Woayra amesiwo léa avu la, elabena woayɔwo be Mawuviwo»

Enye Togo yeyea fofo avafiawodzifia Gnassingbe Eyadema fe didi vevi be amesiame nanɔ dzidzɔ kple ŋutifafa me lle yefe Denyigba la katã dzi, Dzre, fuéle kple mamãwo manɔ amewo kple habɔbɔwo domo o. Eyaŋuti dukplɔla Eyadema anukwaretɔ la lɔna ɖɔɖɔɖowo wɔwɔ edziedzi le dukɔmeviwo dome ɖo.

Ete ŋu dze edzi be miafe tatɔ nutefewɔla la do go kple mia nɔvi Mɔslem Hame fe amegãwo nyitsɔ le RPT fea me hena wodome nyawo sese,

Tototɔ kple dzre ɖo Mɔslem Hamea fe dɔdzikpɔha la me etefe didi. Ale mamã kple fuléle ɖo wo dome.

Le wofe anyinɔnɔa kple numeɖeɖeawo me na To o yeyea fofo vɔ megbe la, miafe ŋutifafa anukwaretɔ Eyadema ɖo asi Hamea fe dɔdzikpɔha la dzi elabenà ŋutif fa kple lɔlɔ̃ mele wo dome o

Eɖe nu me na Xɔsetɔawo be ɖekawɔwɔ kple ŋutifafa dim yele le yefe anyigba la dzi..,

Mlɔeba la dukplɔla la ɖo dɔléɖeasihaa ɖe, si me ame 17 le la na mia nɔviawo bena woadzra Mɔslemtɔwo fe takpekpe gã. si wɔ ge woala la le Lome le Dasiãmime 4 lia dzi la fe dɔwɔwɔwo ɖo,

Avafiawo Menveyinoyu Dzafalo, dũdɔnunɔla Lãmesẽ kple Hadomegbɛnɔnya-wogbɔkpɔla kple Ayaovi Asila, Togo tuakɔwo fe tatɔ kpeɖeŋutɔ hekpe ɖe dukplɔla fe dɔwɔfe dzikpɔla afɛtɔ Kpotivi Têvi Dzidzagbe Lacle ŋu woenɔ tatɔ Eyadema ŋu le Mɔslem Hamea fe nyawodɔdrɔ̃ me. Eye afɛtɔ Mama Fuseni kple Kasim Mensa, siwo nye Hamea fe bubu 'mewo la hã nɔ Takpekpea me nyitsɔ ma,

Slide 16

Innovation 2 / 3

Two layers: a server and a skill

MCP server · the plumbing
  • A standardised, read-only interface any AI assistant can call — Claude, ChatGPT, or open-source
  • Both collections behind one door; the data stays on our servers
AI skill · the intelligence
  • A curator's method, written down — search strategies, transliteration variants
  • Bias and gaps disclosed in every answer

Slide 17

Innovation 2 · the potential

Grounded answers

Built — and where it's going The IWAC server already runs, open source: github.com/fmadore/iwac-mcp-server. The goal: fold in the Eisener collection, then a public assistant on the project website, on an open-source model.

Slide 18

From a plain question to cited sources

The IWAC MCP server installed as an extension in Claude Desktop — read-only, ~22 tools, no API key for the core tools

A Claude Desktop extension — read-only, ~22 tools.

Question in natural language — the AI runs the tools

  • Scopes the collection, then searches it in French
  • Reads the strongest hits in full
  • Claims linked back to its IWAC record

Live for the IWAC; the same door onto the Eisener estate is the goal.

Slide 19

Brief — the default depth

claude.ai/share/d37cbcb6… Full chat ↗
A Claude conversation: asked in English what the Islam West Africa Collection says about secularism in Côte d'Ivoire, the iwac-mcp skill reads its method files and then offers a Brief-or-Extended depth choice, defaulting to Brief

Claude Sonnet 4.6 · high effort

Slide 20

Extended — the full method

claude.ai/share/2bdf2009… Full chat ↗
A Claude conversation: asked about Islam in Kpalimé, Togo, the user answers ‘extended’, and the iwac-mcp skill begins its five-phase analysis with Phase 1 scoping

A five-phase pass — every claim traced to a record in the collection.

Slide 21

Innovation 3 / 3

Equitable, accessible, sustainable

Designed to run anywhere, and to be checked by people.

  • Open-source models by default; commercial AI only as fallback
  • Runs locally — but requires some infrastructure
  • Human-in-the-loop: experts validate before anything enters the database
  • All code open access — reuse and adapt the pipelines: fmadore/iwac-ai-pipelines ↗
The AI-NER-Validator interface: a French newspaper text with AI-extracted people, places, and organisations colour-highlighted, and a sidebar where an expert confirms or corrects each entity before it enters the database.
AI-NER-Validator — experts verify AI-extracted keywords before entry.

Slide 22

Potential downsides

Embracing and managing risk

RiskMitigation
AI invents text (hallucination) or modernises historical spelling Original scans preserved beside every AI output; all models and prompts documented on GitHub
The chatbot approach is largely untested on these collections Iterative development; we document what works and what does not
Collections are only partially catalogued AI complements — never replaces — manual cataloguing
AI tools evolve rapidly Modular architecture; swap components as better tools emerge

Slide 23

Where it leads

Building sustainable networks

A concluding workshop turns the tools into a lasting network

Slide 24

Thank you

Comparing the “peripheries”

AI-driven DH lets two locked archives speak to each other and questions the centre they were measured against.