Dev Translation

IT Technical Translation

Objectives

• Translate technical documents accurately while preserving meaning
• Maintain consistency of technical terminology
• Create bilingual documentation with proper formatting
• Handle code blocks, formulas, and technical diagrams appropriately

Core Translation Rules

Technical Term Handling

On first use, add English in parentheses:

• "智能体 (agent)" not just "智能体"
• Use standard Chinese translations
• Maintain consistency throughout document

Never translate:

• Variable names, function names, class names
• Code blocks and syntax
• Mathematical formulas
• URLs, file paths, commands
• Configuration keys

Always translate:

• Explanatory text and descriptions
• Section headings and titles
• Image captions
• Error messages

For comprehensive term database: See references/terminology-database.md

Bilingual Format (Recommended)

Use sequential format for readability:

## Introduction

This is an introduction to the agent system.

这是对智能体 (agent) 系统的介绍。

---

## Getting Started

Follow these steps to begin.

按照以下步骤开始。

Quality Requirements

Before finalizing:

• Technical terms consistent
• Code blocks unchanged
• Formulas preserved
• Chinese reads naturally (not word-for-word)
• Terminology matches industry standards

Translation Workflow

1. Analyze structure - Identify sections, code blocks, technical terms
2. Create glossary - Build consistent term mapping for this document
3. Translate section by section - Preserve structure, add Chinese below English
4. Validate - Check consistency, formatting, and readability

Using LLM APIs

Recommended Free/Cheap APIs

• Gemini API: 1500 requests/day free, good quality
• DeepSeek API: ~$0.001/request, excellent Chinese
• Groq API: Free tier, ultra-fast

Translation Prompt Template

prompt = f"""请将以下英文技术文档翻译成中文。要求：

1. 保持技术术语的准确性
2. 专业术语后面用括号标注英文，如：智能体 (agent)
3. 翻译要流畅自然，符合中文表达习惯
4. 不要翻译代码、公式、变量名
5. 保持 Markdown 格式

英文原文：
{english_text}

中文翻译："""

Batch Translation Pattern

def translate_markdown(input_path, output_path):
    content = read_file(input_path)
    sections = split_by_markers(content)

    for section in sections:
        chinese = call_llm_api(section)
        append_translation(section, chinese)
        time.sleep(4)  # Rate limiting

    write_file(output_path, translated_content)

Common Patterns

Academic Paper

## Abstract | 摘要

English text...
中文翻译...

API Documentation

## `getUserById(id)` | 根据 ID 获取用户

Returns user info. | 返回用户信息。

Tutorial

## Installation | 安装

Install using npm: | 使用 npm 安装：
\`\`\`bash
npm install package
\`\`\`

Common Mistakes

❌ Translating code:

# Bad: 定义 函数(参数)
# Good: def function(parameter)

❌ Inconsistent terms:

Bad: 智能体, 代理, agent (mixed)
Good: 智能体 (agent) (consistent)

❌ English grammar in Chinese:

Bad: 这个函数返回一个 user 对象
Good: 这个函数返回一个用户 (user) 对象

References

For detailed examples: See references/translation-examples.md For comprehensive IT terms: See references/terminology-database.md For API integration: See references/api-integration.md