CivicSignal
CivicSignal · MediaCloud Ingestion · Analysis · Continental Scale

The infrastructure behind every article CivicSignal sees.

MediaCloud is the ingestion and analysis layer underneath CivicSignal — the discovery pipeline that pulls articles from thousands of African digital sources, and the research tools (Explorer and Topic Mapper) that turn that archive into evidence: attention over time, competing narratives, and networks of influence.

Built on Media Cloud, the open-source platform originally developed at Harvard's Berkman Klein Center and incubated at the MIT Media Lab — the same academic infrastructure used by newsroom and academic researchers worldwide, run here as CivicSignal's own instance for Africa.

56.9M+stories in CivicSignal's MediaCloud archive
56source collections curated for African media
2,000+media sources crawled daily
14,400+new stories added daily
What it is

Two research tools, one continuously growing archive.

MediaCloud runs the discovery-and-extraction cascade — RSS, sitemap, homepage fallback — that pulls articles from thousands of sources into a shared archive. On top of that archive sit two research tools built for exactly this kind of investigation: Explorer, for a fast first read on how an issue is covered, and Topic Mapper, for going deep on one story until the network of who's citing whom becomes visible.

Scale

Continental coverage

Thousands of digital sources monitored across dozens of countries, organised into curated collections, added to daily rather than pulled in one-off snapshots.

Languages

English, French, Arabic

Search and analysis support across CivicSignal's three most-covered languages, with more African-language coverage growing as the archive does.

Output

The shared archive

Every other CivicSignal tool — Wavelength, Fingerprints, Major Events — reads from what MediaCloud collects. It's the foundation layer, not a separate product.

Tool 01 · Explorer

A fast first read on how a story is covered.

Use Explorer to search half a billion stories from more than 50,000 sources including online news and blogs — individually or as curated collections of Africa's top media sources — and get a quick overview of attention, language, and the entities involved.

CivicSignal MediaCloud Explorer tool interface

Track attention over time

See how much an issue is covered, week by week. Spikes reveal the events that drove a conversation; plateaus reveal the "normal" baseline of attention to measure everything else against. Every chart and its underlying aggregated data is downloadable.

Attention-over-time line chart from CivicSignal MediaCloud Explorer, showing coverage volume across outlets of different political leanings Real CivicSignal MediaCloud Explorer output — explorer.civicsignal.africa

Identify how an issue is talked about

Which words do outlets actually use to describe an issue? Word clouds, word counts, bi-grams, word trees and word embeddings surface the differing narratives at a glance — before anyone has to read a single article.

Word cloud output from CivicSignal MediaCloud Explorer showing frequently used terms in coverage of an issue Real CivicSignal MediaCloud Explorer output — explorer.civicsignal.africa

Map geographic coverage

Every story is geocoded to the countries and places it's actually about. That makes it possible to see exactly where an issue is being covered most — and just as usefully, to spot the "issue deserts" where it isn't being talked about at all.

World map showing geographic distribution of story coverage in CivicSignal MediaCloud Explorer Real CivicSignal MediaCloud Explorer output — explorer.civicsignal.africa
Tool 02 · Topic Mapper

Go deep on one story until the network becomes visible.

Once Explorer has narrowed a query down to the sources and time period worth investigating, Topic Mapper turns that starting point into a proper research corpus — collecting more stories, measuring influence, and slicing the results into comparable subtopics.

CivicSignal MediaCloud Topic Mapper tool interface

Discover more content automatically

A Topic starts seeded with stories already in the archive. From there, a web crawler follows the links inside those stories to find more matching coverage — a process called spidering, repeated 15 times to build out a bigger corpus, with rigorous de-duplication of stories that appear at more than one URL.

Topic Mapper dashboard showing total stories, media sources, story links and media links, plus a sentence-volume chart Real CivicSignal MediaCloud Topic Mapper output — topics.civicsignal.africa

Measure influence, not just volume

Every story in a Topic is processed for inlinks and outlinks, so citation and cross-linking between sources becomes measurable — alongside Facebook share counts per story, as a second, independent read on influence. Full network maps are downloadable for further analysis in tools like Gephi.

Table of top stories in a Topic Mapper topic, with media inlinks, outlinks and Facebook share counts per story Real CivicSignal MediaCloud Topic Mapper output — topics.civicsignal.africa

Slice and dice with subtopics

Split a Topic into subtopics for real comparative analysis — automatically by week or month for time-based or key-event analysis, or by a growing list of techniques: boolean keyword queries, country of focus, audience partnership, dominant themes, or media type.

Topic Mapper's subtopic-creation screen, showing techniques including keyword search, audience partnership, top countries, top themes and media type Real CivicSignal MediaCloud Topic Mapper output — topics.civicsignal.africa
Why it matters for CivicSignal

The foundation everything else on this site is built on.

Wavelength's streamgraph, Fingerprints' publisher profiles, Major Events' clustering — none of it exists without MediaCloud's ingestion running underneath. It's the least visible part of CivicSignal, and the part that makes everything visible possible.

56.9M+total stories archived
56curated source collections
2Ksources crawled every day
14.4Kstories added every day
Research partners. CivicSignal's MediaCloud analysis draws on the same open infrastructure used by academic and newsroom researchers worldwide, with research partners including the Reuters Institute for Journalism and the Global Disinformation Index.
Sources

Tool descriptions and screenshots drawn directly from CivicSignal's own MediaCloud deployment — explorer.civicsignal.africa and topics.civicsignal.africa — plus mediacloud.org's description of the underlying open-source platform. Live stats (stories archived, collections, daily crawl volume) captured at time of writing and not auto-refreshed on this page.

The map is moving. CivicSignal is how you keep up with it.

Investigate the full spectrum of Africa's media landscape — from legacy broadcasters to digital-native newsrooms, creator-journalists, and the newsletters and channels that never touch a website.

Connect to CivicSignal
CivicSignal · Code for Africa ← All CivicSignal pages