On Tuesday, 27 October 2026, Vice Rector for Academic Affairs of Universitas Syiah Kuala (USK), Prof. Dr. Heru Pahlevi, met Dr. William Tjhi, Deputy Director of AI Products at AI Singapore, after he held a fruitful discussion with the Data Science and Artificial Intelligence Research Center (DSAI-RC) team at Gedung D, 2nd Floor, FMIPA, USK. Prof. Heru welcomed the visit and expressed hope that the collaboration with AI Singapore would continue and lead to real AI innovation in many fields.
The meeting at DSAI-RC began at 14.00 WIB. It brought together the DSAI-RC team, guests from the Aceh Provincial Language Office (Balai Bahasa Provinsi Aceh), colleagues from Yayasan Ekosisten Lestari (YEL), and students in USK’s Master’s program in Artificial Intelligence. DSAI-RC head, Prof. Dr. Taufik Fuadi Abidin, S.Si., M.Tech., opened the meeting, welcomed Dr. William to the research center, and introduced the team. After that, Khairil Ilmi presented AcehX, a collection of texts, datasets, and language models for natural language processing in Acehnese. Acehnese is spoken by millions of people, yet it remains a language with few digital resources.
AcehX began with a corpus of 63,367 Acehnese sentences. The team used it to build AcehXBERT, an IndoBERT model whose vocabulary was extended with Acehnese words and then retrained on the AcehX corpus. The work was published at O-COCOSDA 2025 as “Development of AcehX for Sentiment Analysis Using a BERT-Based Model”, and in 2026 a new dataset for spelling correction was added. The texts come from six sources, chosen to reflect Acehnese in both written and spoken forms: literary works, folk tales, oral traditions, customary law manuscripts, Wikipedia, and YouTube comments.
The labeled dataset holds 20,000 training, 5,000 validation, and 5,000 test examples. It supports sentiment analysis, spelling correction, normalization of old manuscript texts, and Acehnese–Indonesian translation. Sentiment analysis and spelling correction are complete. Work in progress includes digitizing hikayat manuscripts, an Acehnese–Indonesian machine translation system, a text inspection application, and Acehnese speech recognition. Live demos followed: a spelling-correction tool that labels the error type in each word and offers an Indonesian translation, and a speech-recognition system that turns Acehnese and Gayo recordings into text.
In the discussion, Dr. William asked how the AcehX data was collected and focused on the text inspection application. Beyond correcting spelling and showing word meanings, it is planned to recognize the four main Acehnese accent groups: Aceh Rayeuk, eastern Aceh, Pidie, and western–southern Aceh. It also aims to trace when words and sentences first appeared, document how the language has changed, and identify a text’s origin.
Prof. Dr. Ir. Hammam Riza, M.Sc., IPU, the initiator of AcehX, a USK professor, and chair of the Indonesian AI collaboration network KORIKA, said the cooperation around AcehX will continue and stressed the need to collect more data from untapped languages. Dr. William then introduced SEA-LION, the large language model AI Singapore develops for Southeast Asian languages, and Project ATLAS, an open-source platform of text, speech, and image datasets that anyone can use and contribute to. Prof. Dr. Essy Harnelly, S.Si., M.Si., explored AI collaboration in genomics, and Prof. Dr. Ir. Roslidar, S.T., M.Sc., IPM, ASEAN Eng., raised AI for healthcare, a main focus for AI Singapore alongside language. Meanwhile, Cut Ida Agustina of Balai Bahasa Provinsi Aceh discussed with Dr. William how regional languages in Aceh and across Indonesia could be mapped.
The meeting ended with the signing of the Implementation Arrangement: Integrating the Aceh-X Dataset into Project ATLAS AI Singapore. Aceh is home to many regional languages, among them Acehnese, Gayo, Alas, Aneuk Jamee, Tamiang Malay, Kluet, and Simeulue. Data for Acehnese and Gayo are already available and will continue to improve through this partnership, while DSAI-RC is now building data for other Acehnese languages.
