AWS Startups Blog
How Do You Say ‘JSTOR’ in Arabic? How ADRI is translating Arabic academic literature at scale
Ali Mazraeh was working toward his masters’ degree in Strategic Studies at Victoria University of Wellington, in New Zealand, when he got a taste for how hard it can be to access academic content in Arabic online. Originally from Iran and resettled in New Zealand by the UNHCR, Mazraeh was working hard to create a new future for himself. He found it incredibly difficult to access the credible academic literature he needed to pursue his studies because so much of Arabic academic work is only available in print. “This was the spark that lit the idea,” he said. He decided he wanted to develop a platform where publishers across the world could share and access academic literature in Arabic at the push of a button. Doing so would require changing the way that academic research is preserved and disseminated across the Middle East and North Africa. But the goal to help Arabic scholars share their knowledge globally was a worthy one.
In 2014, Mazraeh founded his company, the Arabic Digital Reform Institute. Over the next two years, ADRI went through an intense cycle of research and development. The first major challenges were technical. ADRI was operating in what’s known as a “greenfield” space—an area of software development where there is no previously existing work to support the project. There was little Arabic academic content online to populate the platform. “A majority of the Arabic academic content remains in hard copy to date,” explained Mazraeh. Finding an efficient way to access and digitize that content was a serious impediment.
Because of these technical challenges, ADRI decided not to reinvent the wheel. Instead, Mazraeh and ADRI ventured into the open-source market to find a way to create an academic repository for scholarly content in Arabic. They found DSpace, a readily available solution that was already widely used for institutional archives. By adopting existing technical architecture and data models, ADRI was able to save time and money.
Still, many linguistic challenges remained. Existing content and document management systems don’t yet accommodate Arabic’s linguistic characteristics, and Arabic translation tools tend to be subpar. After two years of trying, ADRI concluded that the bespoke solution for translating Arabic texts wasn’t feasible with their resources. It was a setback, but it “helped to build resilience among the team,” Mazraeh said. Instead, ADRI decided to leverage its access to Amazon’s translation services to create an automated translation engine and integrate it into its platform. Now, the company uses artificial intelligence and Amazon Translate’s natural language processing to translate from English to Arabic instantly Bulk Arabic translation, of this kind, is unique and ADRI is now a leader in automated translation.
ADRI’s services and vision are in high demand. The company has already established a network of international partners that will ensure they’re able to deliver their services throughout Arabic-speaking communities. They have attracted partners to undertake the quality check of ADRI’s translated literature from English to Arabic. In early 2017, ADRI also signed a partnership with the International Telecommunication Union (Arab States) to deliver its service to 14 universities in the least-developed countries of the Middle East and North Africa. The partnership is scheduled to start in early 2019 through to 2023.
ADRI is a truly international company. Based in Wellington, New Zealand, its members work in Cairo, Hamburg, and London, and its core team speaks 12 different languages. “ADRI has a strong sandbox culture where we encourage experimentation and diversity of thought,” said Mazraeh. The goal is to help Arabic scholars share their work globally. In doing so, ADRI hopes to prove itself to be a leading innovator in the EdTech sector in the Middle East.