ARXIV_SUBMISSION.md

arXiv Submission Checklist — data2skills

Paper Metadata

Title: data2skills: Extracting Interpretable Expert Knowledge
       via Gradient-Optimized Text Skills

Author: Yiwei Xu

Affiliation:
  Institute of Applied Physics and Materials Engineering (IAPME)
  University of Macau
  &
  Songshan Lake Materials Laboratory
  Dongguan, China

Abstract: [see paper/main.tex — already polished]

Category Selection

Primary:   cs.LG  (Machine Learning)
Secondary: stat.ML (Machine Learning - Statistics)
           cs.AI  (Artificial Intelligence) — optional

Rationale: The paper contributes a new learning paradigm (text-gradient descent) with strong connections to interpretable ML and optimization. cs.LG is the natural primary. stat.ML covers the statistical comparison methodology.

Submission Steps

1. Go to https://arxiv.org/submit
2. Click "Start New Submission"
3. Upload the compiled PDF from GitHub Actions
4. Fill in metadata as above
5. Choose license: arXiv.org perpetual, non-exclusive (standard) OR CC BY 4.0 (recommended for maximum visibility)
6. Submit

Pre-Submission Checks

• All figures/tables have captions
• All references are real papers (verified)
• No TBD or TODO markers
• Author affiliation is correct
• GitHub repo linked
• PDF compiles without errors (GitHub Actions)
• Final PDF downloaded and visually checked
• Author email confirmed (replace yiwei.xu@example.com)

After arXiv Acceptance

• Post on X/Twitter with link
• Update GitHub README with arXiv badge
• Submit to Papers With Code
• Consider submitting to a workshop (NeurIPS XAI, ICML workshops)

What to Change in main.tex Before Final Submit

1. Replace yiwei.xu@example.com with real email
2. Consider adding an ORCID if you have one
3. Double-check the abstract is exactly what you want

License Recommendation

CC BY 4.0 — allows others to share and adapt with attribution. Standard in ML community. Your code is already MIT.