Academic Paper To Chinese Insight PDF

Academic Paper to Chinese Insight PDF

Produce a readable Chinese-first paper deliverable, not a raw dump.

Default output contract

Unless the user explicitly asks for literal full translation, prefer a Chinese insight edition with these parts:

1. paper basics
2. what the paper is about
3. why it matters
4. method summary
5. key experiments/results
6. your insights and opinions
7. plain-language explanation
8. concise takeaways
9. polished Chinese PDF export

This default is usually more useful than a noisy page-by-page translation.

When to choose literal translation vs insight edition

Choose literal/full translation when

• the user explicitly asks for full translation
• the paper is short enough that fidelity matters more than readability
• the user needs section-by-section academic preservation

Choose insight edition when

• the extracted text is noisy or layout-corrupted
• the user wants understanding,观点,启发, or explain-like-I’m-younger output
• the user asks for a PDF handout/report

When you choose insight edition, say briefly that you are using a faithful Chinese精读整理 instead of brittle mechanical translation because it improves readability.

Workflow

1. Extract text from the PDF.
2. Inspect extraction quality before deciding output mode.
3. Identify core sections: abstract, introduction, method, results, conclusion, limits.
4. Draft the Chinese report in Markdown.
5. Add your own insights as clearly labeled analysis, not as quoted paper claims.
6. Add a plain-language explanation section when useful.
7. Export with the Chinese PDF renderer.
8. Verify the PDF exists and has non-trivial size.

Files to use

• Use scripts/extract_paper_text.py to extract the paper text into Markdown.
• Use scripts/render_cn_pdf.py to export the final Chinese Markdown into a polished PDF.
• Read references/output-structure.md when shaping the final report.
• Read references/quality-bar.md when deciding translation depth and delivery quality.

Practical guidance

• Prefer semantic cleanup over preserving broken OCR/layout artifacts.
• Keep paper facts separate from your commentary.
• Preserve benchmark names, model names, dataset names, and metric names in original English when that avoids ambiguity.
• If extraction is partially corrupted, rely on recoverable sections and clearly avoid pretending you saw details you could not verify.
• For long papers, summarize tables/results into the most decision-useful findings instead of reproducing every number.
• For Chinese output, optimize for readability first, literalness second, unless the user requests otherwise.

Suggested final filename pattern

paper-topic_论文精读_中文版_YYYY-MM-DD.pdf

Avoid vague names like final.pdf.