AI-Based System to Accommodate Arabic Language and Its Varieties for Natural Language Processing Applications

Tuesday - October 24 2023, 11:01 UTC - 1 year ago

A group of researchers and engineers from the University of Sharjah have developed a deep learning system to utilize the Arabic language and its varieties in applications related to Natural Language Processing (NLP), with a focus on addressing the limitations NLPs encounter with languages of the right-to-left script such as Arabic. Once launched, the system is expected to improve performance for applications such as machine translation, sentiment analysis, and speech recognition, thereby contributing to cultural preservation, accessibility, and more effective cross-cultural communication.

A group of researchers and engineers from the University of Sharjah have developed a deep learning system to utilize the Arabic language and its varieties in applications related to Natural Language Processing (NLP), an interdisciplinary subfield of linguistics, computer science, and artificial intelligence.The scientists say their project will introduce major improvements to NLP systems to accommodate the Arabic language and its dialects when programming computers to process and analyze large amounts of natural language data and assist in developing programs to enhance different language learning skills and boost translation accuracy.

The group, which includes academics and engineers, embarked on the project to assess the usability and usefulness of the Arabic language for AI-powered applications to help nearly half a billion Arabic speakers in the world to benefit from current trends in AI technologies. The results of their work have appeared in international journals.

The new AI-based system which the scientists are creating addresses the limitations NLPs encounter when processing languages other than English. The problem exacerbates with languages like Arabic whose right-to-left script and diacritics, which computers normally fail to recognize, hugely diverge from languages based on the Latin Alphabet.

To address the issue, Dr. Ashraf Elnagar, professor of computer sciences at the University of Sharjah in the United Arab Emirates, has been leading a team of academics to develop a series of computational tools that will assist programmers with the identification of not only formal Arabic but its various dialectal texts.

"The successful completion of the project has the potential to be widely adopted by the masses, as it offers numerous benefits and improvements to various AI-driven language applications and services," says Dr. Elnagar. "It has the potential to cater to a diverse range of users and industries, promoting more effective communication, accessibility, and localization." .

Elaborating on the system, Dr. Elnagar says once launched, it will improve performance and user experience of applications such as machine translation, sentiment analysis, and speech recognition to accurately identify not only the standard Arabic but its numerous dialects, thereby contributing to cultural preservation, accessibility, and more effective cross-cultural communication.

Improving the status of the Arabic language with the aid of AI has become an urgent matter in Arabic speaking countries of the Middle East where computer-savvy users have started leaning on ChatGPT and other AI-driven applications to quickly generate information, execute writing assignments and improve other language skills.

Dr. Elnagar says the project draws on student research at both undergraduate and graduate levels. The project rooted in the Department of Computer Science at the University of Sharjah, showcases the remarkable talents and dedication of our students. Its inception was as a senior project by undergraduate students," notes Dr. Elnagar.

"Later, another student expanded [the] work, using it as the basis for his thesis, with a funder from Sharjah Research Academy," Dr. Elnagar concluded.