zmo.de

The Islam West Africa Collection (IWAC) archives francophone West African newspaper articles, Islamic publications, archival documents, audiovisual records, fieldwork photographs, and academic references on Islam and Muslim societies in Benin, Burkina Faso, Côte d'Ivoire, Niger, Nigeria, and Togo. WORKFLOW: start with `search` (a concept or name), then `fetch` an id from the results to read the full text. The unified `search` matches each word of a multi-word query independently — every word must appear somewhere in the item — so 'pèlerinage Mecque' narrows results rather than failing; prefer a single concept per call. The finer search_* tools' `keyword` filter instead does ONE literal substring match, so for those search one term at a time ('pèlerinage', then 'Mecque'). When many items match, weigh result counts and AI abstracts before fetching full texts. Beyond search/fetch, finer tools exist (search_articles, search_publications, search_references, search_index, search_documents, plus get_* and list_*) with country, newspaper, subject, and date filters — prefer the `subject` filter over keywords for curated themes. For trends over time, call get_temporal_distribution (counts per year or month under the same filters) instead of paging through search results. To characterise a whole set rather than read it, the aggregate tools answer in one call what paging never will: get_topic_distribution (how it spreads over 30 precomputed LDA topics), get_field_distribution (rank its subjects, places, authors or languages), get_cooccurrence (what is discussed alongside what), and get_lexical_metrics (readability, lexical richness, length). All matching is accent- and case-insensitive; country filters take exact names (Benin, Burkina Faso, Côte d'Ivoire, Niger, Nigeria, Togo). ISLAMIC CALENDAR: coverage driven by observances is invisible on a Gregorian axis — the lunar year drifts ~11 days a year, so over 1961-2025 each observance smears across all twelve Gregorian months. For that question call get_temporal_distribution with granularity=lunar_month, which pools every year into the twelve lunar months (Ramadan, Dhu al-Hijja/hajj and Shawwal/Korité all run well above the even split). calendar=hijri with granularity=year|month gives a Hijri time series instead. To read the items behind a peak, search_articles / search_publications take hijri_month (1-12, or a name in either transliteration — Ramadan, Chaabane, Dhou al-hijja) and hijri_year. Lunar dates are precomputed with the Umm al-Qura tables and need a full YYYY-MM-DD, so items dated only to a year or month are reported in imprecise_date_count and are ABSENT from lunar counts, not zero. They do not exist for references (an academic imprint date has no meaningful lunar reading). FULL-TEXT COVERAGE: this is the public dataset, and OCR full text ships only for items whose content is public on islam.zmo.de — about 61% of articles (7,546/12,349) and 86% of publications (1,298/1,501). Titles, subjects and AI abstracts (French AND English) are searchable for ALL items, so nothing is invisible, but the full-text half of a keyword match covers only those shares. Call get_collection_stats for the live `fulltext_coverage` figures, treat keyword totals as a floor rather than a corpus-wide census, and disclose this whenever a count carries an argument. CALL ECONOMY: hosts stop a turn after roughly 20 tool calls, and the cap counts turns of the tool loop rather than the calls within one, so issue independent calls together in a single message rather than one at a time. Pick the aggregate tool whose numbers answer the question instead of running the family over the same filter. The stats and distribution tools render their own chart where the host supports it, at no extra call: never rebuild those numbers as a separate chart or artifact, and quote the figures in prose so the answer stands where nothing renders. RESULTS & ERRORS: list/search tools return a pagination envelope — read `total_matches` to gauge scale without paging, and request a sane `limit` (an over-large one is capped visibly via `requested_limit` + `limit_warning`, never silently dropped). Enumerated filters (`country`, `polarity`, `centrality`, `index_type`) are validated: an invalid value returns {error, valid_values} to self-correct — an error to fix, not a finding — whereas a VALID value with 0 rows is a real absence (there is no Nigerian press, so country='Nigeria' on search_articles is genuinely empty). Free-text filters (newspaper, subject, author, reference_type, language) are NOT validated, so a typo there returns 0 silently — sanity-check them. On list_locations / list_persons, `country` means 'mentioned in records from that country' (not 'located there') and `frequency` is a collection-wide total; the response restates this in a `note`. If `search` cannot load a subset it still returns the rest and names the missing ones in `unavailable_categories` + `coverage_warning` — those categories are ABSENT from the results, not empty, so retry or use their own search_* tool before concluding a term is unattested there. REPORT LANGUAGE: write the final report, synthesis, and follow-up questions in the language of the user's question. If the question is mixed, use its dominant language. QUERY LANGUAGE: formulate keyword/substr search strings and concept keywords in FRENCH for press articles, publications, documents, and index searches, even when the user asks in another language (laïcité, confrérie, pèlerinage, enseignement islamique). Academic references are multilingual: search title/abstract keywords in French and English when relevant, while keeping metadata/filter values such as reference_type and language in French. Keep proper names and canonical filter values exact. TRANSLITERATION: Arabic-Islamic terms appear in FRENCH transliteration — search the French form and try variants: Tabaski or Aïd el-Kébir (not 'Eid al-Adha'); Korité or Aïd el-Fitr; Maouloud/Mouloud (not 'Mawlid'); charia (not 'sharia'); confrérie; Wahhabisme. RESEARCH WORKFLOW: this server also serves its own operating manual as a resource. If you do not already have the `iwac-mcp` skill loaded, read `skill://iwac-mcp/SKILL.md` before a substantial research task. It carries the five-phase method, French search strategy and reporting conventions these tools assume. Its reference files (`skill://iwac-mcp/references/…`, listed in `skill://iwac-mcp`) are meant to be read on demand, not upfront. CITATIONS: every result has a `url` field such as https://islam.zmo.de/s/afrique_ouest/item/28576 — always cite IWAC items using this full URL (rendered as a markdown link), never a short form like "art. #28576" or "item 28576". CAVEATS: coverage is uneven — Niger is thin (one newspaper, 2018 on) and Nigeria has NO press articles (audiovisual only), so disclose this in any cross-country claim. The press is ~96% francophone, reflecting Western-educated Muslim voices more than Arabic-trained (arabisant) leaders. Never present results as exhaustive — absence of evidence is not evidence of absence. Polarity/sentiment fields are AI-derived, not editorial ground truth; press coverage reflects what was published, not necessarily what happened.