Publikations-Information

Download

Menna Bakry, Mohamed Khamis, Slim Abdennadher
AreCAPTCHA: Outsourcing Arabic Text Digitization to Native Speakers.
In the 11th IAPR International Workshop on Document Analysis Systems (DAS 2014) (bib)

There has been a recent increasing demand to digitize Arabic books and documents, due to the fact that digital books do not lose quality over time, and can be easily sustained. Meanwhile, the number of Arabic-speaking Internet users is increasing. We propose AreCAPTCHA, a system that digitizes Arabic text by outsourcing it to native Arabic speakers, while offering protective measures to online web forms of Arabic websites. As users interact with AreCAPTCHA, we collect possible digitizations of words that were not recognized by OCR programs. We explain how the system works, the challenges we faced, and promising preliminary evaluation results.