Digitizing Bhutanese Documents and Books using Machine Translation

The Innotech Department of the royal government of Bhutan partnered with Omdena to conduct a two-month innovation project aimed at developing a machine translation model. The model was designed to automatically translate documents written in Bhutan’s local language to English and store them in the cloud as digital copies. The project was successfully completed by a global team of 50 AI engineers working collaboratively.

Starting with an ideathon in Bhutan

Druk Holding & Investments, the commercial arm of the Royal Government of Bhutan, collaborated with Omdena to organize a groundbreaking virtual ideathon. The event provided Machine Learning and Data Science enthusiasts and experts with a platform to submit their technology-driven ideas to tackle various social challenges, including climate change, healthcare, and energy, among others, in Bhutan.

The primary objective of the ideathon was to identify the most impactful idea that could bring about a positive change in Bhutan through an innovation project. Akash Phaniteja Nellutla emerged as the winner with his outstanding solution, which then got implemented in the following eight weeks innovation project with 50 collaborators.

The impact

Bhutan boasts eighteen different languages, with Dzongkha being the only one with a native literary tradition. Digitizing and translating Bhutanese literature into English would promote its globalization and preservation, while also benefiting from English’s flexibility, versatility, and richness.

Bhutan’s development has been founded on the principle of gross national happiness, with a strong focus on economic growth and development. The country’s initiative to digitize texts and documents related to cultural heritage and literature can drive the development of its digital cultural heritage, creative industries, and IT.

With over two thousand temples and monasteries, both government and private, believed to house significant texts, documents, and manuscripts, digitization of such materials would further encourage the growth of digital cultural heritage, creative industries, and IT in Bhutan.

The project outcomes

Deployed app using Streamlit

Over eight weeks, starting from August to October 2022, a team of 50 AI engineers from all over the world, including some of whom are Bhutanese, developed a machine translation model that automatically translates words as well as sentences from Dzongkha to English. Using various NLP techniques, the model also adds meaning to a complete sentence to make sure translations are logical in sentence structure.

The Omdena and DHI project’s results will support the development of Bhutan’s digital culture, creative industries, and IT.