Jump to content

SciWiki:Assisted article creation

From SciWiki
Revision as of 16:23, 30 October 2026 by NewWiki (talk | contribs)
AI-Assisted Article Creation Guide

SciWiki Science Article Development & Review Internship

Help build trusted science knowledge in Indian languages. Review, enrich, and translate science articles for the SciWiki community.

Internship Languages

Telugu తెలుగు Hindi हिन्दी Tamil தமிழ் Bengali বাংলা Kannada ಕನ್ನಡ Gujarati ગુજરાતી Marathi मराठी

Pilot project

Overview

As a SciWiki intern, you will review, enrich, and translate science articles for sciwiki.net in the seven pilot languages above. This guide sets out the standard procedures and prompt templates for using AI assistance effectively while keeping strict factual integrity. AI tools can speed up first drafts, but every article is built from supplied source material and finished by human review. Regardless of the scientific domain, all interns are expected to perform these fundamental tasks:
Drafting, Translation & Structure
Create or translate comprehensive lead introductions, structure articles into clear sections, build infoboxes, and ensure precise transliteration of scientific terminology.
Quality, Formatting & Metadata
Run manual quality checks to eliminate machine-translation errors. Format text to Wikipedia markup standards, insert inline citations, add interlanguage links, and assign appropriate categories.
Follow these five steps for every AI-assisted article:
STEP 1
Gather Sources
Collect the images, PDFs, text files, or wikitext you will work from. No source means no article.
STEP 2
Prompt the AI
Fill in the Master Prompt Template with your subject and target language, then attach the sources.
STEP 3
Verify Line by Line
Check every sentence, number, and citation against the source. Remove anything you cannot trace.
STEP 4
Polish the Language
Rewrite awkward literal phrasing, confirm terminology, and fix transliteration by hand.
STEP 5
Format & Publish
Add categories and interlanguage links, preview the page, and run the final checklist below.

Core Operational Guidelines

1. The Source Gatekeeper (Strict)
  • Required input: Work exclusively from provided source materials: images, PDFs, text files, or pasted MediaWiki wikitext. If no source is provided, do not generate an article.
  • Zero external knowledge: Rely entirely on the source. Do not fill gaps, add outside background context, or introduce unverified data.
2. Wikitext Translation & Preservation Rules

When working with existing wiki markup, preserve its structure:

  • Images & media: Keep all original filenames (for example, [[File:Example.jpg|...]]). Translate only the captions.
  • References: Keep all original <ref> tags, URLs, authors, and publication titles intact.
  • Categories & links: Keep the original conceptual categories and internal links, translated accurately into the target language (for example, [[Category:Physics]] becomes its target-language equivalent).
  • Templates & markup: Do not remove, merge, or reorder templates, tables, or formatting unless the target language requires it.
3. Zero Hallucination & Traceability
  • 1-to-1 mapping: Every sentence, fact, number, and citation must be directly traceable to the source.
  • No fabricated data: Never invent statistics, URLs, authors, or publication details.
  • Be transparent about gaps: If text is unreadable or information is missing, flag it with an editorial comment such as <!-- Source unclear: figure in table 2 not legible --> instead of guessing.
4. Editorial & Language Standards
  • Native fluency: Use natural, professional phrasing in the target language and avoid awkward literal translations.
  • Scientific terminology: Use established regional scientific terms and precise transliteration. Where no standard term exists, give the English term in brackets on first use.
  • Neutral tone: Write objectively, in encyclopedic prose, without promotional or opinionated language.
  • Infobox standards: Build standard HTML-fallback wiki-tables suitable for SciWiki, with all visible labels and captions translated into the target language.

Standardized Master Prompt Template

Copy and adapt the template below when starting AI-assisted article creation or translation. Replace the items in square brackets, then attach your source files.
Role: You are an expert science encyclopedia writer, translator, and editor for SciWiki (sciwiki.net), fluent in English and highly proficient in Indic languages.

Task: Analyze the attached source material regarding [Enter Subject Here] and generate a complete, well-organized wiki-markup article in [Target Language, e.g., Telugu].

Source Rules (strict):
- Use ONLY the attached source material. If no source is attached, stop and say so; do not write an article.
- Do not add external knowledge, background context, or unverified data.
- Every sentence, fact, number, and citation must be traceable to the source.
- Do not invent statistics, URLs, authors, or publication details.
- If text is unreadable or information is missing, insert an editorial comment (<!-- ... -->) instead of guessing.

Quality Standards:
1. Length & Depth: Aim for comprehensive, encyclopedic prose divided into clear sections using standard wiki headings (e.g., == వివరణ ==, == వర్గీకరణ ==).
2. Neutrality & Tone: Maintain an objective, scientific tone using established regional terminology.
3. Citations: Support factual claims with simple inline tags (<ref>[URL Page_Title], Publisher, Date.</ref>) and include a references section (<references />).
4. Infobox: Build the infobox using standard wiki-table markup, translating labels into the target language using data strictly from the source.

Wikitext Preservation (when the source is wikitext):
- Keep all File/image names unchanged; translate only captions.
- Keep all <ref> tags, URLs, authors, and publication titles intact.
- Keep categories and internal links, translated into the target language.

Output: Return only the finished wikitext, followed by a short list of any editorial comments or unresolved gaps.

Additional Prompt Templates

Translating Existing Wikitext
Translate the pasted wikitext into [Target Language]. Preserve all markup, File names, <ref> tags, URLs, and templates exactly. Translate only prose, captions, table labels, and category/link names. Do not add or remove facts. Mark anything unclear with <!-- ... -->.
Fact-Check Against Source
Compare the draft article below with the attached source. List every sentence, number, name, or citation that is NOT directly supported by the source, and every source fact that was changed or omitted. Do not rewrite the article; only report discrepancies.
Language & Terminology Review
Review this [Target Language] text for unnatural literal translation, inconsistent scientific terminology, and incorrect transliteration. Suggest fixes in a table (original, suggestion, reason). Do not change the meaning or add new information.

Pre-Submission Checklist

  • ☐ Every statement and number is traceable to the provided source.
  • ☐ No external facts, invented URLs, authors, or publication details remain.
  • ☐ Image and media filenames are unchanged; only captions are translated.
  • ☐ All original references are present, and a <references /> section is included.
  • ☐ Scientific terms and transliterations are consistent and checked by a human reader.
  • ☐ The infobox labels are translated and its data comes only from the source.
  • ☐ Categories, internal links, and interlanguage links are added and render correctly.
  • ☐ The lead section summarizes the article, and the tone is neutral and encyclopedic.
  • ☐ All editorial comments are resolved or intentionally left for reviewers.
  • ☐ The page has been previewed for broken markup before saving.

Target Science Domains

You will apply these editorial standards across a wide variety of scientific fields. Choose topics that align with your expertise or the project's current priorities:
Life Sciences & Biodiversity
❖
Species documentation: Translate taxonomic descriptions, morphology, and habitats for endemic Indian flora and fauna.
❖
Genetics & Cell Biology: Explain molecular processes, evolutionary biology, and biotechnology breakthroughs.
❖
Conservation: Document ecosystems, biodiversity hotspots (e.g., Western Ghats), and endangered statuses.
Physics & Space Science
❖
Fundamental Physics: Standardize terminology for quantum mechanics, thermodynamics, and optics.
❖
Astrophysics: Translate complex concepts in cosmology and planetary science.
❖
Indian Space Research: Build comprehensive pages on ISRO missions, satellites, and astronomical observatories.
Health & Biomedical Sciences
❖
Public Health: Create accessible content on epidemiology, nutrition, and tropical medicine.
❖
Human Anatomy: Provide accurate anatomical and physiological descriptions in Indic languages.
❖
Pharmacology: Document medical diagnostics, drug mechanisms, and evidence-based virology.
Chemistry & Materials
❖
Chemical Sciences: Structure articles on physical, organic, and analytical chemistry.
❖
Materials Science: Cover polymers, nanotech, and modern semiconductor manufacturing.
❖
Historical Metallurgy: Document India's metallurgical heritage, such as Wootz steel and traditional zinc distillation.
Earth & Environmental
❖
Climate Science: Develop localized content on the South Asian monsoon, climate change, and meteorology.
❖
Geology & Hydrology: Explain soil sciences, mineralogy, and regional groundwater hydrology.
❖
Disaster Resilience: Structure articles on disaster preparedness, flood management, and drought resilience.
Agriculture & Food Sciences
❖
Agronomy: Document crop science, plant breeding, and horticulture adapted to Indian soils.
❖
Sustainable Farming: Cover bio-pesticides, soil fertility, and traditional water management networks.
❖
Food Systems: Create articles detailing food technology, dairy science, and agricultural biotechnology.
Tech, AI & Engineering
❖
Emerging Tech: Translate complex concepts in Artificial Intelligence, machine learning, and robotics.
❖
Core Engineering: Enhance coverage of civil, mechanical, and electrical engineering principles.
❖
Energy: Build entries on renewable energy systems, green hydrogen, and battery storage.
Mathematics & Biographies
❖
Data Sciences: Write clear explanations of statistics, calculus, geometry, and algorithms.
❖
Scientists & Innovators: Build detailed biographical infoboxes and life histories for Indian researchers.
❖
History of Science: Document the timeline of scientific institutions and landmark discoveries.

Note on AI and Machine Translation: AI tools may help produce initial drafts, but your primary role is human curation. Do not blindly accept machine output. You are responsible for ensuring that the final science articles are factually accurate, faithful to the source, culturally contextualized, and linguistically natural. Never paste AI output into a live article without reading and verifying every line.