Methodology
Terra-Orb is a global media monitoring & analysis engine. Its operating rules are public and stable, so that anyone can verify what the application shows — and what it does not do.
Global neutrality
The 222 countries and territories covered are treated under identical rules: the same detection thresholds, the same topic families, the same barometers. No country is favoured in the ranking, and the Global Front Page deliberately caps each country's representation to avoid any centralisation.
Public data collection only
Terra-Orb only reads public content: RSS/Atom feeds, sitemaps, public pages, public social media accounts, structured data. No paywall bypassing, no classified data, no closed databases. robots.txt directives are respected.
Explainable scores
Every ranked article carries the reason for its score: freshness, pick-up by other media, source reliability, intensity, reach, and topic clarity. Every early signal shows its deviation from normal, its spread speed and its confidence level. Nothing is a black box.
Freshness and deduplication
Only publications from the last 24 hours are retained. Articles covering the same information are grouped into a single event: the wall displays the original and counts the pick-ups, without duplicating.
Limited role of artificial intelligence
Ranking, dating, geography and signals rely on verifiable rules, not on a generative model. AI only steps in to translate headlines and write summaries based solely on aggregated figures — never to invent facts.
Acknowledged limitations
Projections at 6, 24 and 72 hours are statistical extrapolations with a confidence level attached, not predictions. Early signals are probabilistic: a signal may not be confirmed, and Terra-Orb then displays it as such.
See also: the source directory, the coverage comparator et the public API.