Patent & research
Patent pending.
The method behind Lebanese.ai is protected by a provisional patent application filed with the United States Patent and Trademark Office by Code-X.
The filing
A dialect AI translation system.
The application covers how Lebanese.ai reads dialect input written in Arabic script, in Arabizi or in a mix, maps it to one form, and returns it in the language the user needs.
- Status
- Patent pending
- Office
- United States Patent and Trademark Office
- Application
- Provisional 64/021,845
- Applicant
- Code-X
The method
What the model does, step by step.
- Step 1
Dialect input
Text or audio arrives as people type it or say it: Arabic script, Arabizi, a mix, or speech.
shu3amta3mel - Step 2
Normalisation
Spellings, scripts and sounds are mapped to one form the model can reason over.
3am→عم - Step 3
Dialect memory
The model applies what it knows about how the dialect is really used.
dialectcontext - Step 4
Output
The answer comes back in dialect, Modern Standard Arabic, English or French, or in all three languages mixed in one sentence.
شو عم تعمل؟
Research
Low-resource dialects, first.
Our research starts from dialects that have little written data and many ways of being written and spoken. Lebanese is the first. Findings are published here once they clear review.
Technical briefings
Researchers, partners and investors can ask for a briefing on the model and the filing.
Ask for a briefing