Weekly Signal
One week of coverage at a time, broken down by topic. Step back with ‹ to compare against earlier weeks, or filter by country and language below; the legend re-ranks itself by whichever week and filter you're looking at.
Major Events
Correlated coverage, not a single keyword hit — each card below is independent reporting from multiple publishers converging on the same real-world incident within a tight date window. Detected automatically from shared entities and timing; not hand-curated, so treat categories and entities as a strong signal, not a verified fact-check.
Search the corpus
Keyword search across profiled publishers and recent coverage — search by name, country, topic or a word from a headline.
Start typing to search across 759 profiled publishers and the latest coverage feed.
Start typing to search the ~600 most-mentioned people, organisations and places across every source type — not just what's on this page today.
CivicSignal can stand up a customized tracking report — your keywords, your countries, your cadence.
Topic & language mix, whole corpus
Share of all classified articles by topic and by detected language, across every publisher and country in this dataset.
Topic mix
Language mix
Language mix reflects article-level detection, not self-declaration — a language can show 0% here and still be one a publisher self-reports covering. A quiet share of the corpus, not confirmed absence.
Recent coverage
Latest classified stories across the corpus — pick up to three topics to compare side by side. Each column's summary is a live snapshot built from the same data below it, not a generated write-up.
If it does, you're looking at something with real room to grow. Digital-native publishers, creator-journalists and Push Media channels are already tracked continuously across the countries CivicSignal covers — daily coverage across every market, earlier detection of emerging narratives, and continent-wide Push Media monitoring are within reach. We'd welcome a conversation with anyone who sees the same potential.
Some notes on methodology
What this is
A working showcase unifying three source types: scraped digital-native publishers and creator-journalists (titles, body text, country and publication date), plus Push Media — newsletters, Telegram and fediverse channels monitored by a separate Code for Africa platform. All three run through the same topic & language classifier.
How topics are assigned
Each article is scored against bilingual (English/French) keyword sets across ten topics; the highest-scoring topic wins. Articles matching nothing are left as "Other" and excluded from the charts above.
Known limits
Publication dates lean heavily on recent scrape runs, keyword classification is heuristic rather than semantic, Arabic-language coverage is detected but not yet topic-classified with dedicated keywords, and Push Media sources are cross-border by nature — they're labeled "Pan-African" rather than attributed to one of the seven tracked countries.