elsevier-paper-miner

name	elsevier-paper-miner
description	Mine the lawful template-paper corpus for section structure, figure/table/caption patterns, and reviewer-risk lessons without copying protected text.
family	elsevier
version	1.2.0
kind	macro
status	experimental-unofficial

Elsevier Paper Miner

Purpose

Mine the local paper_templates/ corpus for manuscript architecture. This skill studies structure, section flow, evidence organization, table/figure/caption logic, and reviewer-risk patterns. It must not copy text, figures, tables, captions, equations, benchmark interpretations, or paper-specific claims.

Corpus sources

• Corpus manifest: paper_templates/corpus_manifest.yaml and paper_templates/corpus_manifest.json
• Paper cards: paper_templates/cards/
• Extracted structure maps: paper_templates/extracted_structure_maps/
• Thematic mining notes: paper_templates/mining_notes/
• Open-access papers bundled in this release: 6
• Official template/manual PDFs bundled in this release: 1

When to use

Use this skill when the user asks for:

• a template-paper based section skeleton;
• abstract / introduction / related work / method / experiment patterns;
• table, figure, or caption style mining;
• reviewer-risk lessons derived from lawful open-access examples;
• journal-specific examples without copying source content.

Required operating rules

1. Use only corpus entries with allowed source/license metadata.
2. Prefer paper_templates/cards/<paper-id>.md before opening the PDF.
3. Use extracted_structure_maps/ for section sequencing and visual-evidence patterns.
4. Use mining_notes/ for cross-paper style summaries.
5. Never copy source text, captions, figures, tables, equations, or paper-specific claims.
6. If the user wants submission/legal guidance, route to elsevier-skill-suite and official-source fragments instead of relying on template papers.
7. For real submission, live-check publisher and target-journal instructions.

Workflow

1. Detect target publisher, journal, article type, and manuscript component.
2. Select the smallest relevant corpus subset.
3. Read the Paper Card first; use the PDF only to confirm structural observations.
4. Extract a neutral skeleton or checklist.
5. Rewrite using the user's own evidence, terminology, and experimental results.
6. Return a copyright-safe output with a source-use warning.

Output contract

Return:

• selected corpus entries;
• source/license status;
• structural observations;
• reusable section skeleton;
• table/figure/caption lessons if relevant;
• reviewer-risk checklist;
• forbidden-copying warning;
• items requiring live official-source verification.

v1.2.0 evaluation benchmark awareness

When demonstrating this skill, prefer the benchmark tasks in evals/tasks/ and sample-output contracts in evals/sample_outputs/. For paper-mining tasks, always include a copyright-safe warning and use template papers only for structure and pattern extraction.

Extracted visual assets

Use paper_templates/extracted_visual_assets/ when the user wants to analyze main figures, motivation/framework diagrams, or table design patterns. These assets are for structure analysis only and must not be copied into manuscripts. Always include the copyright-safe warning and point to visual_asset_manifest.json when citing the corpus internally.