Case studies
Public Services & Religion · Vietnam
AI-powered Buddhist Scripture Translation System for Long Hung Pagoda
A web application for managing and translating Buddhist scriptures (the Tripitaka) for Long Hung Pagoda, using AI fine-tuned for Pali, Sanskrit and Classical Chinese plus a chatbot that suggests writing style, shortening translation time compared with traditional methods.
JavaJavaScriptPythonJHipsterAngularSpring BootMongoDBKubernetesAI fine-tuningChatbot
Project overview
- Client: Long Hung Pagoda – the Buddhist Book Translation and Publishing Centre under the Vietnam Buddhist Sangha, one of the largest centres of Buddhist study in Hanoi
- Country: Vietnam
- Industry: Religion – translation of Buddhist scriptures
- Scale: 10.71 MM, 3-month development period; system built for around 50 users
- Project type: Web Application
- Scope of work: SRS, Basic design, Detailed design, System Architecture design, Coding, Testing, Maintenance
Problem / Challenges
The client needed to optimise the process of translating and managing Buddhist scriptures, applying AI to improve translation quality. The initial plan was to translate the Tripitaka from Classical Chinese, Sanskrit and Pali, then extend to other Buddhist texts. Users needed a chatbot to support both translation and style reference.
Challenges:
- The required languages (Pali, Sanskrit) are very rarely used, so the project team had to build the solution from scratch.
- The style of Buddhist scriptures differs greatly from the everyday language used to train AI models.
Solution
- Integrated a chatbot into the translation interface, letting users translate with AI while receiving suggestions to adjust the style.
- Researched and experimented with fine-tuning pre-trained models to reduce the amount of data needed while still improving translation quality.
- Fine-tuned the system multiple times to adapt to the style of Buddhist scriptures.
Key features
- Processing and reading content from Word, Excel and PowerPoint documents.
- Managing members in the system and in each translation project.
- Managing documents per project.
- File translation with two modes: quick and advanced.
- AI-assisted multilingual translation: Sanskrit, Pali, Chinese, English, Vietnamese.
- Saving translated content as a dictionary for future reference.
- A chatbot that supports users during translation.
Technology
- Languages: Java, JavaScript, Python
- Framework/Platform: JHipster, Angular, Spring Boot
- Database: MongoDB
- Infrastructure: Kubernetes
Results
- Built a centralised file management and translation tool for Long Hung Pagoda.
- Shortened translation time compared with traditional methods, as AI helps users reach the desired result as quickly as possible.
Lessons learned
- Setting up an automated deployment process from the very start of the project reduces deployment time and makes the system more professional.
Have a similar project?
Contact SSD