open-research 1.1.2 → 1.2.0

package/builtin-skills/draft-paper/references/venues/chi.md

@@ -0,0 +1,131 @@
+# CHI (ACM Conference on Human Factors in Computing Systems) Submission Guidelines
+
+## LaTeX Template
+
+Use the ACM sigconf template. Replace the generic preamble with:
+
+```latex
+\documentclass[manuscript,review,anonymous]{acmart}
+% For camera-ready (accepted): \documentclass[sigconf]{acmart}
+
+\usepackage{booktabs}
+\usepackage{graphicx}
+\usepackage{amsmath}
+
+% ACM-specific metadata (fill in for camera-ready)
+% \acmConference[CHI '26]{CHI Conference on Human Factors in Computing Systems}{April 2026}{Yokohama, Japan}
+% \acmDOI{...}
+% \acmISBN{...}
+
+\bibliographystyle{ACM-Reference-Format}
+```
+
+For submission (review), use the `manuscript,review,anonymous` options which produce single-column, double-blind output.
+
+## Format Constraints
+
+- **Layout**: Single-column for review; final publication is 2-column (ACM handles conversion via TAPS)
+- **Length**: Measured by word count (NOT pages), excluding references, captions, and appendices:
+  - Short paper: ≤5,000 words
+  - Standard paper: 5,001–12,000 words (typical: 7,000–8,000)
+  - Over 12,000 words triggers desk rejection unless strongly justified
+- **No fixed page limit** — length governed by word count
+- **Template**: Must use official ACM template. Wrong template = desk rejection.
+- **Bibliography style**: ACM-Reference-Format (numbered citations [1], [2])
+- **Figures**: Vector formats preferred (PDF, EPS, SVG). Include \Description{} for alt text.
+
+## Required Sections
+
+CHI does not mandate rigid section names, but reviewers expect:
+
+### Empirical/Study Papers
+Abstract → Introduction → Related Work → Method/Participants → Analysis/Results → Discussion → Conclusion → References
+
+### System/Design Papers
+Abstract → Introduction → Related Work → Design Process → System Architecture → Evaluation → Discussion → Conclusion → References
+
+### What Each Section Must Accomplish
+- **Introduction**: Establish the HCI problem, motivate why it matters, state contribution EXPLICITLY. Don't make reviewers guess.
+- **Related Work**: Must be comprehensive enough to contextualize in HCI scholarship. Grossly insufficient lit review = Assisted Desk Rejection (ADR-Context).
+- **Method**: Sufficient detail for replication. Explain WHY, not just WHAT. Missing this = ADR-Method.
+- **Results/Findings**: Adequate data to support claims. Insufficient data = ADR-Data.
+- **Discussion**: Critically engage with results, limitations, generalizability, ethics. Shallow discussions are flagged.
+
+## Anonymization (Double-Blind)
+
+- Remove author names, affiliations, acknowledgments, document metadata
+- DO NOT anonymize references — cite your prior work normally
+- DO write about your own work in third person: "Smith et al. [10] showed..." NOT "Our previous work [10]..."
+- Marking any reference as "[Anonymous]" = desk rejection
+- All supplementary materials must be anonymized
+- Preprints (arXiv) allowed but CHI notes they can create unconscious reviewer bias
+
+## Abstract
+
+- **Maximum**: 150 words
+- Must not be a placeholder — incomplete metadata triggers desk rejection
+
+## Accessibility Requirements (Important for CHI)
+
+- **Alt text**: Provide meaningful \Description{} for every figure
+- **Color**: Don't convey info through color alone — add patterns, shapes, line styles
+- **Tables**: Use actual table markup, not images
+- **Equations**: Use LaTeX math, not images
+- **Video figures**: Must include closed captions (.srt or .sbv)
+- Inaccessible papers may be reassigned to non-optimal reviewers
+
+## Review Criteria
+
+### Five Core Dimensions
+1. **Significance** — Why does this contribution matter to HCI?
+2. **Originality** — What new ideas, methods, systems, or knowledge?
+3. **Research Quality** — How rigorous, transparent, well-executed?
+4. **Presentation Clarity** — Concise, well-organized, jargon-free?
+5. **Prior Work Engagement** — Adequately reviewed and critically engaged?
+
+### Length-to-Contribution Proportionality
+Reviewers assess whether paper length is commensurate with contribution. Long papers must justify their length through substantive contribution.
+
+### Assisted Desk Rejection (ADR) Criteria (New for CHI 2026)
+1. **ADR-Context**: Severely inadequate literature review
+2. **ADR-Method**: Insufficient methodological transparency
+3. **ADR-Data**: Grossly inadequate data for claims
+4. **ADR-Contribution**: Disproportionately small HCI contribution for length
+
+### Review Process
+CHI uses a Revise & Resubmit model:
+- Round 1: Threshold decisions. Papers with ≥1 "Revise & Resubmit or higher" advance.
+- Round 2: Final "Accept with minor revisions" or "Reject." No further revision.
+
+## Common Rejection Reasons
+
+**Desk rejection triggers:**
+- Wrong template or format
+- Any anonymization violation (paper, supplements, external links)
+- Marking any reference as "anonymous"
+- Over 12,000 words without justification
+- Missing or placeholder abstract/metadata
+- Unmarked LLM-generated content
+
+**Substantive rejection:**
+- Paper length incommensurate with contribution
+- Insufficient research quality or rigor
+- Inadequate theoretical grounding
+- Unclear or unjustified design choices
+- Not explaining the "why" behind methodological choices
+- Shallow discussion of ethics, limitations, generalizability
+
+## Camera-Ready Specifics
+
+- Switch to \documentclass[sigconf]{acmart}
+- Add CCS concepts (generate at dl.acm.org/ccs/ccs.cfm)
+- Add keywords (4-8 terms)
+- Include acknowledgments and author info
+- Retain ACM copyright/DOI block without alteration
+
+## Key Policy Notes
+
+- Papers reviewed "mostly on an as-is basis" — submit in near-publishable form
+- Make contribution explicit and early
+- LLM use for beyond editing must be disclosed; unmarked = desk rejection
+- Acceptance rate: ~25%

package/builtin-skills/draft-paper/references/venues/ieee-transactions.md

@@ -0,0 +1,173 @@
+# IEEE Transactions Submission Guidelines
+
+## LaTeX Template
+
+Use the IEEEtran document class. Replace the generic preamble with:
+
+```latex
+\documentclass[journal]{IEEEtran}
+% For Computer Society journals (TPAMI, TSE): \documentclass[10pt,journal,compsoc]{IEEEtran}
+% For Communications Society: \documentclass[journal,comsoc]{IEEEtran}
+% For technical notes/letters (9pt): \documentclass[9pt,technote]{IEEEtran}
+
+\usepackage[T1]{fontenc}
+\usepackage[utf8]{inputenc}
+\usepackage{amsmath, amssymb}
+\usepackage{graphicx}
+\usepackage{booktabs}
+\usepackage{cite}
+\usepackage{hyperref}
+\usepackage{microtype}
+
+% Restore page breaks within equations when using amsmath
+\interdisplaylinepenalty=2500
+```
+
+Template files: template-selector.ieee.org or Overleaf (preloaded). Do NOT modify IEEEtran.cls.
+
+## Format Constraints
+
+- **Paper size**: US Letter (8.5" x 11")
+- **Layout**: Two-column, single-spaced
+- **Font**: Times New Roman, 10pt body
+- **Column width**: 3.5 inches (88.9 mm) single column; 7.16 inches (182 mm) full page span
+
+### Page Limits (Vary by Journal)
+| Journal | Regular Paper | Short/Brief |
+|---------|--------------|-------------|
+| TPAMI (Computer Society) | 14 pages | 8 pages |
+| TNNLS (Comp. Intelligence) | ~14 pages | 3 pages |
+| Signal Processing Society | 13 initial / 16 revised | 2 pages |
+| T-ASE (Robotics & Automation) | 12 pages | 6 pages |
+| Trans. Automatic Control | 12 pages (13-16 mandatory charge) | — |
+
+All limits include references and author biographies. Check specific journal for exact limits.
+
+## Required Sections (Standard Structure)
+
+1. **Title** — Title case, concise, reflective of content
+2. **Author names and affiliations** (omitted in double-blind review)
+3. **Abstract** — 150-250 words, single paragraph
+4. **Index Terms** — After abstract, from IEEE Thesaurus
+5. **I. Introduction** — Use \IEEEPARstart for decorative drop cap
+6. **Body sections** — Roman numeral headings (II, III, IV, ...)
+7. **Conclusion** — Singular; do not replicate abstract
+8. **Acknowledgment** — Singular (not "Acknowledgements"); before References
+9. **References** — Numbered list in order of appearance
+10. **Author Biographies** — With head-and-shoulders photograph; after References
+
+### Section Heading Hierarchy
+- **Primary**: Roman numeral, centered, 10pt bold small caps — "II. RELATED WORK"
+- **Secondary**: Letter label, flush left, italic — "A. Feature Extraction"
+- **Tertiary**: Arabic numeral with close paren, italic, indented — "1) Convolutional Layers:"
+
+## Abstract Requirements
+
+- **150-250 words**, single paragraph ONLY
+- NO abbreviations (except IEEE, SI, ac, dc), footnotes, references, display equations, or tables
+- Self-contained: problem statement, methodology, key results, main conclusions
+- Set in 9pt in published layout (handled by IEEEtran)
+
+## Index Terms / Keywords
+
+- Placed immediately after abstract
+- 4-6 terms from the **IEEE Thesaurus** (~12,420 controlled vocabulary terms)
+- Format: "Index Terms—" followed by comma-separated terms in alphabetical order
+- Used for IEEE Xplore discoverability
+
+## Math and Equations
+
+- Number all displayed equations consecutively: (1), (2), (3)... flush right
+- Appendix equations: (A1), (A2)
+- Refer to equations as "(1)" in text — NOT "Eq. (1)" except at sentence start
+- **Scalars**: italic (*x*). **Vectors**: bold (**x**). **Matrices**: bold uppercase (**A**).
+- Functions: roman (sin, exp, log, max). Units: roman (Hz, km).
+- All symbols defined on first use
+- Use \begin{equation}, \begin{align} — NEVER use eqnarray or $$
+
+## Figure Requirements
+
+- **Resolution**: Min 300 DPI photos/grayscale; min 600 DPI line art
+- **Formats**: PS, EPS, TIFF, PDF, PNG
+- **Single column**: 3.5" wide. **Double column**: 7.16" wide.
+- **Fonts**: Embed all fonts. Use Times New Roman, Helvetica, Arial. 8-10pt labels.
+- **Captions**: BELOW figure. Format: "Fig. 1. Title" with sentence case.
+- **Subfigures**: (a), (b), (c) in 8pt Times, centered below each.
+- **Color**: Verify readability in grayscale — many journals print B&W.
+
+## Table Requirements
+
+- **Numbering**: Roman numerals — TABLE I, TABLE II
+- **Caption**: ABOVE table (opposite of figures). Full caps header.
+- **Formatting**: Vertical lines optional. Use booktabs for clean formatting.
+- **Footnotes**: Letter superscripts (a, b, c), not numbers
+
+## Reference Format (IEEE Style)
+
+Numbered citations in square brackets: [1], [2], [3]. Order by first appearance.
+
+Templates:
+```
+[1] A. B. Author, "Article title," Abbrev. Journal, vol. X, no. Y, pp. 123-456, Mon. Year.
+[2] J. K. Author, "Paper title," in Proc. Abbrev. Conf., Location, Mon. Year, pp. 123-456.
+[3] A. B. Author, Book Title, Xth ed. City, State: Publisher, Year.
+```
+
+- Author format: First initial(s) + family name (J. K. Smith)
+- Journal titles: italic, standard IEEE abbreviations
+- Use standard month abbreviations: Jan., Feb., Mar., etc.
+
+## Review Criteria
+
+### Core Requirements
+1. **Novelty**: New or innovative methods/approaches
+2. **Appropriateness**: Within journal scope, well-written, complete
+3. **Quality**: Technical accuracy and rigor
+4. **Impact**: Significant contribution to field and readership
+
+### Review Decisions
+- **A (Accepted)**: Only minor fixes (typos)
+- **AQ (Accepted with Mandatory Revisions)**: Minor fixes verifiable by AE
+- **RQ (Major Revision)**: Serious flaws requiring second review
+- **R (Rejected)**: Fails novelty/appropriateness
+
+### Review Process
+- Minimum 2 independent reviewers
+- Most journals: single-blind (reviewers anonymous, authors visible)
+- Some journals double-blind: TNNLS, T-ASE, IEEE Communications Letters
+- Plagiarism screening automated before and during review
+
+### Conference Paper Extensions
+- Must cite prior conference publication
+- ≥60-70% new material required (max 30-40% text overlap)
+- Novel additions: expanded theory, more extensive experiments, new analysis
+
+## Common Rejection Reasons
+
+**Desk rejection:**
+- Out of scope
+- Plagiarism >30-40% similarity
+- Dual submission
+- Wrong format (not IEEE two-column)
+- Page limit exceeded
+
+**After review:**
+- Insufficient novelty (incremental variation of prior work)
+- Weak experimental validation (missing ablations, weak baselines)
+- Poor presentation quality
+- Inadequate literature review (<10-15 peer-reviewed citations)
+- Overreliance on conference papers/websites instead of journal references
+- Figures unreadable in grayscale
+
+## Overlength Page Charges
+
+- **Voluntary**: $110/page within standard limit (optional)
+- **Mandatory overlength**: $220-250/page exceeding journal limit (non-negotiable)
+- Open access APC (~$2,800) is separate from overlength charges
+- IEEE member discount: 5-20% on APC
+
+## Key Policy Notes
+
+- Initial submission can be single-column for readability (check journal preference)
+- Camera-ready must be in IEEE two-column format
+- Author biographies with photo required for final version (some journals no longer require photo)

package/builtin-skills/draft-paper/references/venues/nature.md

@@ -0,0 +1,167 @@
+# Nature Journal Submission Guidelines
+
+## LaTeX Template
+
+Nature accepts LaTeX submissions. Use a clean article class — Nature does NOT have a public .sty file like conferences. The editorial team reformats accepted manuscripts.
+
+```latex
+\documentclass[12pt]{article}
+\usepackage[utf8]{inputenc}
+\usepackage[T1]{fontenc}
+\usepackage{lmodern}
+\usepackage[margin=1in]{geometry}
+\usepackage{amsmath, amssymb}
+\usepackage{graphicx}
+\usepackage{booktabs}
+\usepackage{hyperref}
+\usepackage{natbib}
+\usepackage{microtype}
+
+\linespread{2.0}  % Double-spacing required for review
+```
+
+Nature uses a numbered (Vancouver) reference style with superscript citations. After acceptance, the production team applies Nature's house style.
+
+## Article Types and Length
+
+**"Letters" were retired in October 2019.** All original research is now "Article."
+
+### Article (Primary Research)
+- **Physical sciences**: ~2,500 words main text + 4 display items (~6 printed pages)
+- **Biological/clinical/social sciences**: ~4,300 words main text + 5-6 display items (~8 printed pages)
+- **Display items**: Max 8 total (figures + tables combined)
+- **Methods**: Separate section, up to ~3,000 words
+- **References**: ~50 in main text (methods-only references exempt from this count)
+
+### Other Types (less common for researchers)
+- Review Article: ~5,000 words, ~150 refs (usually commissioned)
+- Correspondence: 300-800 words, ~10 refs
+- Matters Arising: <1,200 words
+
+## Document Structure (Canonical Order)
+
+1. **Title** — ~75 characters max, clear to broad readership, no abbreviations
+2. **Authors and affiliations**
+3. **Bold summary paragraph** (the "abstract") — ~200 words, fully referenced, unstructured
+4. **Main text** — NO headed "Introduction" section; weave intro into narrative flow
+5. **Main references** — numbered consecutively by order of appearance
+6. **Tables**
+7. **Figure legends** — listed sequentially after references
+8. **Methods section** — placed AFTER references (not inline), includes:
+   - Data availability statement (mandatory)
+   - Code availability statement (mandatory if custom code used)
+9. **Methods references** — continuing numbering from main references
+10. **Acknowledgements**
+11. **Funding statement**
+12. **Author contributions** (CRediT format)
+13. **Competing interests declaration** (mandatory even if none)
+14. **Extended Data figure/table legends**
+
+## Abstract ("Summary Paragraph")
+
+- ~200 words or fewer
+- Written in **bold**
+- Fully referenced (references appear in main reference list)
+- Unstructured — NO subheadings
+- Must be accessible to readers outside the immediate subfield
+- Avoid numbers, abbreviations, acronyms unless essential
+- Functions as both a non-technical intro AND brief summary of results + implications
+
+## Methods Section
+
+- Placed AFTER main references and figure legends (not inline with main text)
+- Up to ~3,000 words
+- Subdivided by short bold headings for specific methods
+- Specific subsections encouraged for: Statistics, Reagents, Animal models
+- Methods-only references do NOT count against the ~50 reference limit
+- Must contain:
+  - **Data availability** statement
+  - **Code availability** statement
+  - For animal studies: named institutional committee, confirmation of guideline compliance
+  - For human subjects: named ethics committee, informed consent confirmation
+
+## Reference Format (Vancouver/Numbered)
+
+In-text: superscript numbers (e.g., `\textsuperscript{1,2}` or `\textsuperscript{1-4}`)
+
+Reference list format:
+```
+Author(s). Article title. Journal Title vol, pages (year).
+```
+- Author: surname first, then initials
+- 6+ authors: first author et al.
+- Journal title: italic, abbreviated
+- Volume: bold
+- Article titles ARE required
+
+Use \begin{thebibliography}{99} with \bibitem, NOT BibTeX (to maintain precise control over formatting).
+
+## Figures
+
+- **Widths**: Single column 89mm, 1.5 column 120-136mm, full page 183mm
+- **Resolution**: Min 300 DPI (450 DPI recommended)
+- **Font in figures**: Arial or Helvetica ONLY, 5-7pt minimum
+- **Panel labels**: lowercase (a, b, c...), 8pt bold
+- **Color**: RGB, colorblind-friendly palette (avoid red-green)
+- **Scale bars**: Required for microscopy
+- **Format**: PDF, EPS, AI, SVG preferred (vector-based)
+
+## Extended Data vs Supplementary Information
+
+- **Extended Data**: Up to 10 multi-panel items. Peer-reviewed, copy-edited, indexed as part of paper. Online only. Additional figures go HERE, not in SI.
+- **Supplementary Information**: Large datasets, detailed methods too lengthy for Methods. Max 30MB/file, 150MB total. NOT for additional figures.
+
+## Author Contributions (CRediT)
+
+Required for ALL articles. Use CRediT taxonomy (14 roles):
+Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing
+
+Format: "J.S.: Conceptualization (lead), Writing - original draft (lead). K.M.: Methodology (lead), Software (lead)."
+
+## Review Process
+
+- **Editorial triage**: ~70% desk-rejected (some sources say >90%). Decision in 3-14 days.
+- **Bar**: "Science that changes how the field thinks" — technically flawless alone is insufficient
+- **Peer review**: 2-3 external reviewers, 4-8 weeks after assignment
+- **Post-review acceptance**: ~40% of papers reaching peer review
+- **Revision periods**: 6-12 months common when new experiments needed
+- **Overall acceptance rate**: <8%
+
+### What triggers editorial interest:
+- Findings that overturn assumptions (not incremental improvements)
+- Broad interdisciplinary appeal
+- Narrative accessible to non-specialists from paragraph one
+- Resolution of long-standing major problems
+
+## Common Rejection Reasons
+
+**At desk (editorial triage):**
+- Incremental advance without conceptual breakthrough (most common)
+- Better fit in specialist journal
+- Written for specialists, not broad readership
+- Leading with methods instead of findings
+- Incomplete work submitted prematurely
+
+**At/after peer review:**
+- Insufficient statistical rigor (sample size, controls, blinding)
+- Conclusions not fully supported by data
+- Ethical compliance issues
+- Inability to release code or data
+- Poor figure quality or unclear data presentation
+
+## Cover Letter
+
+Must include:
+- Statement that manuscript is original and not under review elsewhere
+- Why findings interest Nature's broad readership
+- The specific assumption the work overturns
+- 4-5 suggested reviewers with affiliations
+
+**Optimal opening**: "We show that X works via Y, overturning the assumption that Z" — not background.
+
+## Presubmission Inquiry (Strongly Recommended)
+
+- 1-page pitch: scientific question, why unresolved, key finding, 1-2 figures
+- Response in ~2 working days
+- ~25% encouraged to submit full paper
+- Positive response ≠ guaranteed peer review

package/builtin-skills/draft-paper/references/venues/neurips.md

@@ -0,0 +1,115 @@
+# NeurIPS Submission Guidelines
+
+## LaTeX Template
+
+Use the official NeurIPS style file. Replace the generic preamble with:
+
+```latex
+\documentclass{article}
+\usepackage[preprint]{neurips_2025}
+% For submission (anonymous + line numbers): \usepackage{neurips_2025}
+% For camera-ready (accepted): \usepackage[final]{neurips_2025}
+
+\usepackage[utf8]{inputenc}
+\usepackage[T1]{fontenc}
+\usepackage{lmodern}
+\usepackage{amsmath, amssymb}
+\usepackage{graphicx}
+\usepackage{booktabs}
+\usepackage{hyperref}
+\usepackage{cleveref}
+\usepackage{microtype}
+\usepackage{natbib}
+```
+
+Only add packages justified by actual content (e.g., \usepackage{algorithm2e} if you have pseudocode).
+
+## Format Constraints
+
+- **Layout**: Single column, US Letter (8.5" x 11")
+- **Font**: Times New Roman, 10pt body, 11pt leading
+- **Title**: 17pt bold, centered, framed by horizontal rules
+- **Page limit**: 9 pages for main content (figures, tables included). References, checklist, and appendices do NOT count.
+- **Camera-ready**: +1 page (10 pages main content)
+- **Max PDF size**: 50 MB
+- **Supplementary**: Single ZIP, max 100 MB
+- **Fonts in PDF**: Must be Type 1 or embedded TrueType only — no bitmap/Type 3
+
+## Required Sections
+
+1. **Abstract** — One paragraph, no explicit word limit but must fit the indented abstract block (~200-250 words practical). Claims must match paper content.
+2. **Introduction** — Problem, why it matters, what others missed, this paper's contribution, roadmap sentence
+3. **Related Work** — Can combine with Introduction or Background
+4. **Method / Approach**
+5. **Experiments / Results** — With error bars, ablations, statistical significance
+6. **Conclusion**
+7. **References** — After main body, unlimited pages
+8. **NeurIPS Paper Checklist** — 16 required items after references (mandatory, desk reject if missing)
+
+### Strongly Recommended
+- **Limitations** section — Reviewers actively look for this. A paper that acknowledges limits is rated higher.
+- **Broader Impacts** — Expected when applicable (malicious uses, bias, privacy, environment)
+- **Acknowledgments** — Use \begin{ack}...\end{ack} (auto-hidden in anonymous mode)
+
+## Anonymization (Double-Blind)
+
+- Remove ALL author names, affiliations, acknowledgments
+- Self-cite in third person: "Jones et al. [4] showed..." NOT "We previously showed [4]"
+- All supplementary materials (code, data, appendices) must be anonymized
+- No identifiable GitHub links, institutional watermarks, or IRB institution names
+- Violation = desk rejection without review
+- Posting non-anonymous preprints on arXiv is allowed and does NOT cause rejection
+
+## Review Criteria (Each Rated 1-4)
+
+1. **Quality** — Claims well-supported by theory or experiments; methodology appropriate; work complete
+2. **Clarity** — Well-organized, enough detail for expert to reproduce results
+3. **Significance** — Impact on community; others likely to use or build on ideas
+4. **Originality** — New insights or understanding; clear differentiation from prior work
+
+### Overall Score (1-6)
+- 6: Strong Accept — technically flawless, groundbreaking
+- 5: Accept — solid, high impact on at least one sub-area
+- 4: Borderline Accept
+- 3: Borderline Reject
+- 2: Reject — technical flaws, weak evaluation
+- 1: Strong Reject
+
+## Reproducibility Checklist (Key Items)
+
+- All assumptions explicitly stated
+- Full experimental details (hyperparameters, splits, optimizer settings)
+- Error bars, confidence intervals, or significance tests for main claims
+- Compute resources reported (GPU type, memory, runtime)
+- Code and data accessible (anonymized links at submission)
+
+## Common Rejection Reasons
+
+**Desk rejection triggers:**
+- Page limit exceeded (>9 pages main content)
+- Modified style file parameters
+- Missing paper checklist
+- Anonymization violations
+- Dual submission
+
+**Scientific rejection:**
+- Weak evaluation: insufficient baselines, missing ablations, no error bars
+- Overclaiming: abstract/intro claims broader than experiments support
+- Missing limitations: if reviewers find limitations you didn't acknowledge, it hurts more than disclosing them
+- Inadequate related work or failure to differentiate from prior work
+- Appendix-heavy papers hiding critical details to circumvent page limits
+
+## Formatting Details
+
+- **Headings**: 12pt bold flush left (1st level), 10pt bold (2nd), 10pt bold inline (3rd)
+- **Math**: Use LaTeX display environments (equation, align), NOT $$ (breaks line numbering)
+- **Tables**: Center; use booktabs (\toprule, \midrule, \bottomrule). No vertical rules. Caption BEFORE table.
+- **Figures**: Caption AFTER figure. Captions lowercase except first word.
+- **References**: natbib loaded by default. Both author/year and numeric styles acceptable.
+- **Footnotes**: Superscript after punctuation. Minimize use.
+
+## Key Policy Notes
+
+- Authors bear full responsibility for AI-generated content including citations
+- Rebuttal: 10,000 characters max, no paper revisions allowed
+- Acceptance rate: ~24-25%