Linguistic Intelligence Matrix.
35 / 35 / 20 / 10.
The definitive linguistic cartography of the Moroccan digital street. Our audience is never monolingual; code-switching is permanent. The engine weights every UGC stream by this real-world volume — and detects the exact moment a Darija bad buzz crosses the membrane into mainstream press.
Why this matrix exists
Centrale Danone lost €150M. The bad buzz started on Facebook (Darija), amplified on WhatsApp (Darija), and crossed into mainstream press (MSA + French) 72 hours later. By then it was too late. The cascade detection catches that crossing in real-time.
Darija is 10% of the volume but 90% of the signal. It's where bad buzz starts — anonymous Hespress comments, TikTok threads, WhatsApp groups. The matrix over-indexes Darija on UGC and excludes it from formal articles (where it was polluting sentiment scores before).
A single 0-100 score weighted by the matrix. A Darija spike alone bumps the score. But when that spike crosses into MSA or French — the cascade alert fires critical. That's the moment the crisis leaves the underground and enters the institutional mainstream.