PRD of Translation Editor

🚀 WG སྡེ་ཚན། 🛠️ AI Tools WG
1
Owning Group Pecha AI studio
Status Complete
GitHub Project Buddhist AI Studio WG · GitHub
Last Updated 2025-09-22

1. Overview

This project is the development of a specialized Computer-Assisted Translation (CAT) tool editor designed specifically for translating Tibetan Buddhist texts into other languages. The tool solves the critical problem of low-quality and inefficient translations in this niche domain by integrating state-of-the-art AI, similar to GitHub Copilot/Cursor, and implementing custom-researched workflows. The primary users are translators, linguists, and scholars working with canonical Tibetan literature who currently lack tools that understand the unique linguistic and semantic complexities of the source material.

2. Goals & Success Metrics

The primary goal is to enhance the quality and accelerate the speed of translating Tibetan Buddhist texts.

  • Goals:

    1. Improve Translation Accuracy: Significantly reduce errors and improve the contextual and terminological accuracy of machine-generated translations.

    2. Increase Translator Efficiency: Decrease the time and effort required for translators to produce a high-quality, publishable translation.

    3. Provide a Superior User Experience: Create an intuitive and powerful editor interface tailored to the specific needs of translating Tibetan script and syntax.

  • Success Metrics:

    • Achieve a 30% reduction in post-editing time compared to baseline CAT tools (e.g., Trados, MemoQ).

    • Increase translation quality scores (COMET/BLEU) by 15% on our internal test sets.

    • Receive a user satisfaction score of 8/10 or higher from a beta testing group of at least 10 translators.

    • Successfully handle 100% of the defined complex workflow test cases.

3. Timeline & Quarterly Milestones

A high-level schedule for the project, broken down by quarter. This aligns with work already completed.

  • Q1 2025 (Completed):

    • Researched and experimented with various AI/ML workflows for Tibetan-to-English translation.

    • Analyzed existing CAT tools to identify strengths and weaknesses for our use case.

  • Q2 2025 (Completed):

    • Designed complete UI/UX mockups in Figma.

    • Planned and finalized the foundational backend data schema.

  • Q3 2025:

    • Milestone 1: Convert Figma mockups into a functional web application (front-end).

    • Milestone 2: Integrate the front-end with core backend services (user authentication, project creation, text segmentation).

  • Q4 2025:

    • Milestone 3: Integrate and test the complex translation workflows researched in Q1.

    • Milestone 4: Implement and refine the AI suggestion engine within the editor interface.

    • Milestone 5: Onboard a closed group of beta testers for initial feedback.

  • Target Launch: Q1 2026 (Public Beta)

4. Scope & Features

  • Included Features:

    • Two-Panel Editor: A clean side-by-side view of the source Tibetan text and the target language translation.

    • AI-Powered Suggestions: In-line, context-aware translation suggestions and completions.

    • Workflow Integration: Support for specialized workflows, such as handling root verses, commentaries, and specific lexical patterns unique to Buddhist texts.

    • Source Text Annotation: Ability for users to view definitions or notes on source Tibetan words.

    • Basic Project Management: Functionality to create, manage, and track the progress of translation projects.

  • Out of Scope (for V1):

    • Real-time multi-user collaborative editing.

    • Advanced project management features (e.g., translator assignment, analytics).

    • Mobile or desktop applications.

    • Support for languages other than Tibetan as the source.

5. Dependencies

  • AI/ML Team: Requires the finalized translation models and APIs to be delivered and maintained.

    • Status: In progress, alpha models are available.

  • Backend Services Team: Depends on the stable deployment of core APIs for user management, database interaction, and text processing.

    • Status: Deployed on staging, integration is ongoing.

  • Tibetan Language Experts & Scholars: Access to subject matter experts for workflow prompt writing, validation and quality assessment of outputs.

    • Status: Partnership established with three scholars for beta testing.

6. Acceptance Criteria

The project will be considered “done” for its V1 launch when:

  • A translator can log in, create a new project, and upload a Tibetan text file.

  • The text is correctly segmented and displayed in the two-panel editor.

  • The AI engine provides relevant, in-line suggestions as the translator types.

  • The translator can accept, reject, or modify AI suggestions.

  • The user can save their progress and export the completed translation.

  • All specialized workflows function as designed on a curated set of 10 test documents.

  • The application passes all end-to-end tests for the core user journey without critical bugs.