A5 Multilingual Translation & Refinement

💡 One-liner: Use AI for translation, refinement, and rewriting while maintaining terminology consistency.

📝 Course Notes

Key takeaways from this lesson:

---

What You'll Be Able to Do

• Translate full documents or sections with AI
• Maintain consistency in technical terminology
• Refine text to improve readability
• Rewrite content to reduce similarity

---

Why OpenCode Instead of Web-based AI?

Capability	Web-based AI	OpenCode
Batch Translation	❌ Copy-paste section by section	✅ Multi-file batch processing, format preserved
Terminology Consistency	❌ AI may translate differently each time	✅ @glossary reference, unified across document
Long Document Handling	❌ Easily exceeds context limits	✅ Section-by-section translation, auto-merge
Format Preservation	❌ Format often lost	✅ Direct file editing, original format retained
Version Comparison	❌	✅ Git tracks translation history

---

Your Current Pain Points

• Inconsistent terminology in translated articles
• Unstable quality for long document translations
• Need to reduce similarity but manual rewriting is exhausting

---

When to Use This Technique

• When you need: High-quality translation or rewriting
• And you don't want to: Spend hours manually adjusting every word

---

🎒 Before You Start

Make sure you've completed the following:

• [ ] Completed A1 Writing Workflow
• [ ] Have text that needs translation or refinement

---

Core Concept

Translation Workflow

Glossary Preparation → Section Translation → Terminology Unification → Final Review

Translation vs Refinement vs Rewriting

Task	Purpose	Typical Scenario
Translation	Cross-language conversion	Translate English paper to Chinese
Refinement	Improve expression quality	Enhance article readability
Rewriting	Change expression, keep content	Plagiarism check, derivative works

---

Follow Along

Step 1: Prepare a Glossary

Why
Consistent terminology is key to professional translation.

Switch to Build mode:

I need to translate a machine learning paper from English.
Help me create a glossary with Chinese translations for these common terms:
- Neural Network
- Deep Learning
- Gradient Descent
- Overfitting
- Regularization

Save as glossary.md

Step 2: Section-by-Section Translation

Why
Breaking long documents into sections maintains quality.

@glossary.md

Please translate the following English paragraph with these requirements:
1. Use standard translations from the glossary
2. Maintain academic writing style
3. Ensure natural, flowing sentences

Original text:
{paste English paragraph}

Step 3: Full Document Translation

Why
Smaller files can be translated in one go.

@glossary.md @paper.txt

Translate paper.txt into Chinese with these requirements:
1. Use standard glossary translations
2. Preserve original paragraph structure
3. Save as paper_cn.md

Step 4: Refine for Readability

Why
Translations usually need refinement.

@paper_cn.md 

Please refine this translation with these requirements:
1. Fix awkward sentences
2. Maintain academic rigor while improving readability
3. Check terminology consistency
4. Edit the file directly

Step 5: Rewrite to Reduce Similarity

Why
Some scenarios require lower similarity scores.

Please rewrite the following paragraph to reduce similarity:
1. Keep the original meaning intact
2. Use different expressions
3. Output 3 versions to choose from

Original text:
{paste paragraph needing rewrite}

---

📋 Magic Prompts

🌐 Professional Multilingual Translation

Expected result: High-quality translation with consistent terminology

## Role
You are a professional translator skilled in making translations natural and fluent while preserving original meaning.

## Task
Translate the user-provided text into the target language.

## Input Information
### Required
- Target language: 【Simplified Chinese / English / Japanese / Spanish / French / German / ...】
- Original text:

[Paste original text]

### Optional
- Glossary: @【glossary file?】
- Domain: 【Technical / Legal / Medical / Literary?】
- Tone style: 【Formal / Casual / Professional?】

## Translation Requirements
1. **Terminology Consistency**: Use standard translations from the glossary
2. **Format Preservation**: Maintain original paragraphs, headings, and list structure
3. **Natural Flow**: Don't translate word-by-word; make the translation read naturally
4. **Proper Nouns**: Add original text in parentheses on first occurrence (e.g., Machine Learning 机器学习)

## Output Format
### Translation
[Translation result]

### Translation Notes (if any)
- Polysemy handling: 【word】translated as【translation】, because【reason】
- Terminology confirmation: 【term】uses【translation】, per glossary

## Constraints
- ✅ Terminology consistency throughout (same term = same translation)
- ✅ Readability over word-for-word faithfulness
- ✅ Use domain-standard translations for technical terms
- ❌ Avoid translationese (awkward phrasing from literal translation)
- ❌ Don't omit original content

## Error Handling
- If target language not provided, ask the user first
- If encountering polysemous words, note possible translations for user to choose
- If a term isn't in the glossary, use common translation and flag it

✏️ Rewriting for Similarity Reduction

Expected result: Multiple rewritten versions maintaining original meaning

## Role
You are a professional editor skilled at changing expressions while preserving meaning.

## Task
Rewrite the user-provided text to reduce similarity with the original.

## Input Information
### Required
- Rewrite level: 【Light (30%) / Medium (50%) / Heavy (70%)】
- Original text:

[Paste original text]

### Optional
- Keywords to preserve: 【terms that must stay?】
- Style requirement: 【Academic / Casual / Formal?】

## Rewrite Strategies
### Light Rewrite (~30%)
- Synonym replacement
- Minor word order adjustments
- Keep core expressions

### Medium Rewrite (~50%)
- Sentence structure changes
- Active/passive voice conversion
- Merge/split sentences

### Heavy Rewrite (~70%)
- Reorganize paragraph logic
- Express same meaning differently
- Add/remove descriptive details

## Output Format
### Version 1
[Rewritten text]

### Version 2
[Alternative rewrite]

### Version 3
[Third rewrite approach]

---

### Rewrite Statistics
| Version | Estimated Change | Main Techniques Used |
|---------|-----------------|---------------------|
| Version 1 | X% | 【technique】 |
| Version 2 | X% | 【technique】 |
| Version 3 | X% | 【technique】 |

**Recommended Version**: Version【N】
**Reason**: 【reason】

## Constraints
- ✅ Preserve original meaning
- ✅ Keep technical terms accurate
- ✅ Make each version distinctly different
- ❌ Don't change core viewpoints
- ❌ Don't introduce information not in the original

---

Checklist ✅

Must complete all before moving on

• [ ] Created a glossary
• [ ] Used @ to reference glossary during translation
• [ ] Completed refinement edits
• [ ] Can generate rewritten versions

---

Common Pitfalls

Issue	Cause	Solution
Inconsistent terminology	No glossary	Create glossary first, @ reference during translation
Quality drops in long texts	Too much at once	Translate section by section, ~2000+ words each
Meaning changed after rewrite	Too aggressive	Specify "preserve original meaning"

---

Lesson Summary

You learned:

1. Translation workflow (glossary → section translation → unification → refinement)
2. Using @ to reference glossaries for consistency
3. Refinement and rewriting techniques

---

Next Lesson Preview

Next, we'll dive into advanced content — Fiction Writing, learning multi-file management, character card design, and maintaining character consistency.

---

📚 More Complete Templates: Prompt Template Library