A Class XII student from Royal Global School made an unusual appearance at an academic conference usually reserved for university researchers, presenting a technical paper on Sanskrit-Chinese phonology at RegICON 2025 held at Gauhati University on Thursday.
Huma Abia Kanta, a student of the science stream, delivered a full-length research paper in a regular session of the Regional International Conference on Natural Language Processing organised by the Department of Information Technology. Her work, co-authored with Dr Ankur Pan Saikia of Assam down town University and Dr Utpal Barman of Assam Skill University, was featured alongside presentations by faculty members and scholars from institutions in India and abroad.
Her study proposes an encoder–decoder model with Luong attention to reconstruct Sanskrit-oriented phonology from Tang-era Chinese Buddhist lexicon entries. The paper is based on a bilingual dataset linking Siddham-derived forms with Chinese characters, and argues that attention-based neural architectures can assist digital philology, historical linguistics and transliteration tasks involving classical Sanskrit and Chinese material.
The session in which her paper appeared also included work on Indian-language benchmarks, Assamese coreference resolution, low-resource machine translation and speech technologies for endangered languages of the Northeast. Conference organisers noted that it is uncommon for school students to be included in these tracks. RegICON 2025 also featured keynote speakers such as Prof Francis Bond and Dr Karma Wangchuk.
Dear Reader,
Every day, our team at EastMojo travels through rain, rough roads, and remote hills to bring you stories that matter – stories from your town, your people, your Northeast.
We do this because we believe in truthful, independent journalism. No big corporate backing, no government pressure – just honest reporting by local journalists who live and breathe the same air you do.
But to keep doing this work, we need you. Your small contribution helps us pay our reporters fairly, reach places others ignore, and keep asking the tough questions.
If you believe the Northeast deserves its own fearless voice, stand with us.
Support independent journalism. Be a Member.
Thank you,
Karma Paljor
Editor-in-Chief, eastmojo.com
The authors detail how the dataset was built, the choices made during tokenisation and model configuration, and the evaluation process used to judge phonetic similarity. They report that such neural models can approximate expert reconstructions with reasonable accuracy, while noting the need for larger corpora, improved handling of dialectal differences and the inclusion of multimodal features in future research.
This is Huma’s second international-level presentation this year, following an earlier AI/ML paper presented in Azerbaijan. Her mentors say her participation reflects the potential of early-stage researchers from Assam and the Northeast when given structured guidance and credible platforms.
Outside her research work, Huma runs “desicodes,” an open-source initiative aimed at making programming more accessible by offering an Assamese-to-Python transpiler called “asPy,” with plans to extend the tool to other Northeastern languages. The project is designed to help first-generation learners and promote regional languages within digital and AI systems.
Huma thanked Royal Global School Chairman Dr A.K. Pansari and School Director Dr Arup Kumar Mukhopadhaya for their support, and credited her mentor Dr Ankur Pan Saikia for encouraging her to explore advanced topics. She said their support motivates her to continue contributing to research and innovation from Assam.
Also read | Kohima street vendors trained in food safety ahead of Hornbill