Loga is a “ReCaptcha-inspired” mobile application for crowd-sourcing Arabic text digitization, which was originally developed at the 2016 NYUAD Hackathon for Social Good where it took first prize and audience choice. The aims of the project are two-fold. Firstly, to increase the availability of Arab digital content in order to preserve Arab literary history. Secondly, to create an employment opportunity for unemployed or marginalized Arabic speakers. The project is currently in development for the Hult Prize, a social entrepeneurship competition, where it has qualified for Semi-finals. The project has garnered attention from many Middle Eastern media outlets such as The National, Arabian Marketer, and Gulf Today.
Metalang can instantly provide a list of all instances when a specific word was said in any collection of audio or video. The system is powered by machine learning with recurrent neural networks, combined with a phonetic algorithm we designed known as the language locality search. The system runs on phonetic alphabet, so it can search for non-dictionary words such as names of people, places, or companies.