management-science-writing — скилл для Claude Code
Скилл management-science-writing помогает писать и редактировать статьи по management science и operations research: от выбора темы и журнала до модели, алгоритма и выводов.
MIT ★ 240 обновлён 2026-09-18
npx skills add franklee16/academic-research-skills --skill management-science-writing -a claude-codeЧто делает management-science-writing
Скилл ведёт работу над статьёй по цепочке «модель → анализ → алгоритм → численный эксперимент → управленческие выводы». Он помогает выбрать журнал, сформулировать модель, оформить математические выводы, описать алгоритм, подготовить эксперименты и связать результаты с практическими решениями.
Поддерживаются ориентиры для MS, OR, MSOM, POM, TS, TRB, DS, OMEGA, TRE, EJOR, IJPE, IJPR и C&IE. Подсказки охватывают выбор темы, структуру статьи, notation, assumptions, proofs, pseudocode и ответы на замечания рецензентов. Для OMEGA текст нужно сверять с ориентиром ~6000 слов.
Какие задачи решает
- выбрать журнал под модель, метод и вклад
- выстроить статью от модели до управленческих выводов
- сформулировать математические доказательства и алгоритм
- подготовить численные эксперименты и анализ чувствительности
- переработать статью по замечаниям рецензентов
Примеры запросов агенту
- Помоги выбрать журнал для статьи с аналитической моделью и алгоритмом.
- Проверь структуру моей статьи по цепочке «модель — анализ — алгоритм — эксперимент — выводы».
- Сформулируй управленческие выводы из этих математических результатов.
Как установить management-science-writing
1Через Skills CLI (Vercel)
npx skills add franklee16/academic-research-skills --skill management-science-writing -a claude-code2Вручную: положить файлы
~/.claude/skills/management-science-writing-main/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/management-science-writing-main/
1Через Skills CLI (Vercel)
npx skills add franklee16/academic-research-skills --skill management-science-writing -a codex2Вручную: положить файлы
.agents/skills/management-science-writing-main/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add franklee16/academic-research-skills --skill management-science-writing -a cursor2Вручную: положить файлы
.cursor/skills/management-science-writing-main/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install franklee16/academic-research-skills paper-writing/management-science-writing-main/SKILL.md2Вручную: положить файлы
.github/skills/management-science-writing-main/SKILL.md1Через Skills CLI (Vercel)
npx skills add franklee16/academic-research-skills --skill management-science-writing -a gemini-cli2Командой платформы
gemini skills install https://github.com/franklee16/academic-research-skills --path paper-writing/management-science-writing-mainВручную: положить файлы
.devin/skills/management-science-writing-main/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Claude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о management-science-writing
Как установить management-science-writing в Claude Code?
Выполните npx skills add franklee16/academic-research-skills --skill management-science-writing-main -a claude-code, затем используйте файл ~/.claude/skills/management-science-writing-main/SKILL.md.
Что делает management-science-writing?
Скилл помогает писать и редактировать статьи по management science и operations research: от выбора темы и журнала до моделирования, доказательств, алгоритмов, численных экспериментов и управленческих выводов.
Для каких журналов подходит management-science-writing?
Он охватывает MS, OR, MSOM, POM, TS, TRB, DS, OMEGA, TRE, EJOR, IJPE, IJPR и C&IE, включая разные требования к методам, практической значимости и формату статьи.
Исходный текст
SKILL.md · MIT
---
name: management-science-writing
description: Use when writing or revising management science & operations research papers targeting MS, OR, MSOM, POM, TS, TRB, DS, OMEGA, TRE, EJOR, IJPE, IJPR, C&IE. Covers the full 0-to-draft pipeline: topic selection, journal targeting, modeling, derivation, algorithm design, numerical experiments, managerial insights, and journal-specific writing patterns. Also use when stuck at any stage of an MS&E paper and needing stage-specific guidance.
---
# Management Science & Operations Research Paper Writing
## Overview
Management Science & Engineering (MS&E) papers follow a distinct pipeline: **Model → Analysis → Algorithm → Numerical Experiment → Managerial Insight**. Unlike ML papers that prioritize empirical results, MS&E papers are evaluated on the **coherence of the analytical chain** — from modeling assumptions through mathematical derivation to computational validation and finally to actionable business implications.
This skill provides journal-specific guidance for 13 flagship MS&E journals across two tiers, covering the full paper lifecycle.
**Core Principle**: An MS&E paper is a **mathematical argument with managerial purpose**. Every equation, lemma, and table must serve the narrative that connects a real problem to a rigorous solution.
---
## When to Use This Skill
- Drafting a paper targeting MS, OR, MSOM, POM, or any MS&E journal
- Deciding which journal best fits a particular model-method-contribution combination
- Structuring a model-analysis-algorithm paper
- Writing proofs, algorithm pseudocode, or numerical experiment sections
- Formulating managerial insights from mathematical results
- Revising based on reviewer feedback from an MS&E journal
- Converting a working paper between MS&E journal formats
---
## Journal Selection Quick Reference
Choosing the right journal is a function of your paper's **methodology depth**, **problem scope**, and **practical relevance**.
### Tier 1 (UTD-24): Top Flagship Journals
| Journal | Core Identity | Ideal Paper Profile |
|---------|--------------|---------------------|
| **MS** (Management Science) | Broad management, high rigor, significant contribution | Novel problem + sharp analytical insight + broad appeal across management disciplines. Length: ~38 pages submission. |
| **OR** (Operations Research) | Methodological depth, optimization focus | Deep theoretical advance in optimization/stochastics. Complete proofs. Methodology-first contribution. |
| **MSOM** (M&SOM) | Empirical OM, behavioral operations | Strong empirical component or behavioral experiment. Managerially relevant. Data-driven modeling. |
| **POM** (Production & Oper. Mgmt) | Data-driven, broad OM scope | Empirical or analytical paper with strong practical relevance to production/service operations. |
### Tier 2: Strong Field Journals
| Journal | Core Identity | Ideal Paper Profile |
|---------|--------------|---------------------|
| **TS** (Transportation Science) | Transportation systems, network models | Transportation-specific problem. Network optimization, traffic equilibrium, logistics. |
| **TRB** (Transportation Research B) | Transportation methodology | Methodological advance in transportation modeling, rigorous proofs. |
| **DS** (Decision Sciences) | Decision-making focus | Decision analysis, multi-criteria methods, judgment under uncertainty. |
| **OMEGA** | Concise, high-impact, managerial | Short, sharp papers (~6000 words max). Strong managerial insights mandatory. |
| **TRE** (Transportation Research E) | Logistics & transportation | Supply chain and logistics modeling. Empirical or analytical. |
| **EJOR** (European J. Operational Research) | Broad methodology, inclusive | Any OR methodology applied to a real problem. Welcomes heuristic methods. Types: Invited Review, Innovative Applications, Theory & Methodology. |
| **IJPE** (Int. J. Production Economics) | Production economics, supply chain | Manufacturing/service operations with economic analysis. Strong on real data. |
| **IJPR** (Int. J. Production Research) | Production engineering | Process design, scheduling, manufacturing systems. Engineering focus. |
| **C&IE** (Computers & Industrial Eng.) | Computational methods, IE | Algorithm-heavy, simulation, AI/ML for industrial engineering. Implementation emphasis. |
### Decision Flowchart
```
Is your contribution primarily methodological?
├── YES → Is it a deep theoretical advance in optimization?
│ ├── YES → OR, possibly MS (Optimization dept)
│ └── NO → Is the method heuristic/computational?
│ ├── YES → EJOR, C&IE, IJPR
│ └── NO → MS, DS
└── NO → Is your contribution primarily empirical/applied?
├── YES → Does it have broad managerial appeal?
│ ├── YES → MS, MSOM, POM, OMEGA
│ └── NO → IJPE, IJPR, TRE, C&IE
└── NO → Hybrid (model + data) → MSOM, POM, EJOR
```
---
## The MS&E Paper Structure
Unlike IMRAD for biomedical sciences or the systems paper blueprint, an MS&E paper typically follows this structure:
| Section | Typical Pages | Key Content |
|---------|--------------|-------------|
| Abstract | 150-250 words | Problem, methodology, key results, managerial implication |
| S1 Introduction | 2-3 pages | Motivation, gap, contributions (numbered), paper organization |
| S2 Literature Review | 2-4 pages | Organized by stream, explicit differentiation table |
| S3 Model | 4-8 pages | Notation, assumptions, base model, extensions |
| S4 Analysis / Derivation | 4-10 pages | Lemmas, propositions, theorems with proofs |
| S5 Algorithm / Solution Method | 2-5 pages | Pseudocode, complexity, approximation guarantees |
| S6 Numerical Experiments | 4-8 pages | Setup, parameter settings, main results, sensitivity analysis |
| S7 Managerial Insights / Discussion | 1-3 pages | Actionable implications, connected to practice |
| S8 Conclusion | 0.5-1 page | Summary, limitations, future work |
| References | — | 30-60 papers typical |
| Appendix | Unlimited | Long proofs, additional experiments, robustness checks |
**Key structural difference from ML/CS papers**: The **Model section is the backbone**. It occupies the most central real estate. If your model is unclear, the rest of the paper collapses. Reviewers read the model section before anything else after the introduction.
---
## Modeling Conventions
See [references/modeling-conventions.md](references/modeling-conventions.md) for comprehensive guidance. Key principles:
### Notation System
- Use a **consistent three-tier notation**: sets (calligraphic), parameters (lowercase), decision variables (uppercase or bold), dual variables (Greek)
- Present a **notation table** (for papers with 15+ symbols)
- Define all symbols at first use; never assume the reader infers from context
### Assumption Presentation
- List assumptions as **numbered statements** (Assumption 1, 2, 3...)
- For each assumption, state: (a) what it is, (b) why it is needed, (c) whether it is restrictive and what happens if relaxed
- **Assumption defense**: Explain why each assumption is reasonable in the target application context
- Use **assumption-relaxation extensions** ($4-6 pages) to demonstrate robustness
### Model Types by Journal Preference
| Model Type | MS | OR | MSOM | POM | EJOR | IJPE |
|-----------|-----|-----|------|-----|------|------|
| Analytical (optimization/game theory) | ✓✓✓ | ✓✓✓ | ✓✓ | ✓ | ✓✓✓ | ✓✓ |
| Stylized analytical + numerical | ✓✓ | ✓✓ | ✓✓✓ | ✓✓ | ✓✓ | ✓✓ |
| Data-driven empirical | ✓✓ | — | ✓✓✓ | ✓✓✓ | ✓ | ✓✓✓ |
| Simulation-based | ✓ | — | ✓ | ✓ | ✓✓ | ✓✓✓ |
| Heuristic/metaheuristic | — | — | — | ✓ | ✓✓✓ | ✓✓✓ |
---
## Proof and Derivation Standards
See [references/proof-derivation-guide.md](references/proof-derivation-guide.md) for comprehensive guidance. Key principles:
### Proof Presentation Hierarchy
```
Lemma 1 (Technical result) → Proposition 1 (Structural property) → Theorem 1 (Main result) → Corollary 1 (Special case)
```
### Journal-Specific Proof Expectations
| Journal | Full Proof in Body | Proof Sketch Tolerated | Appendix Proofs | Proof Length Tolerance |
|---------|-------------------|----------------------|-----------------|----------------------|
| MS | Short proofs in body; long ones appendix | For non-central results only | Yes, unlimited | Concise proofs preferred |
| OR | Complete proofs in body | Rarely | Yes | No length constraint on proofs |
| MSOM | Streamlined in body; full in appendix | Yes, for some results | Yes | Condensed preferred |
| POM | Streamlined in body | Yes | Yes | Condensed preferred |
| EJOR | Complete in body or appendix | Yes | Yes | Moderate |
| IJPE/IJPR | Minimal in body | Yes, often | Yes | Short |
| C&IE | Minimal | Yes, frequently | Optional | Very short |
### Proof Writing Guidelines
1. **State the result before proving it.** Use "We next establish that..." or "The following proposition characterizes..."
2. **Begin each proof with the proof technique** ("By induction on N", "We construct a coupling...", "By contradiction, assume...")
3. **End each proof with a clear marker** (∎ or □)
4. **Connect proofs to intuition**: Before a long proof, give a one-sentence intuitive explanation
5. **All proofs in the main body must be self-contained**; do not reference appendix derivations from body proofs
---
## Algorithm and Solution Method Presentation
See [references/algorithm-solving-guide.md](references/algorithm-solving-guide.md) for comprehensive guidance.
### Pseudocode Standards
- Use the `algorithm2e` or `algorithmicx` LaTeX packages
- Every algorithm must have: input, output, and numbered steps
- Include **computational complexity** (worst-case time and space) for every algorithm
- If applicable, state **approximation ratio** and **tightness**
### Algorithm vs. Heuristic
- **Algorithm**: Provable optimality or approximation guarantee → label as "Algorithm"
- **Heuristic**: No theoretical guarantee, validated numerically → label as "Heuristic" or "Solution Procedure"
- C&IE and IJPR accept pure heuristics; MS/OR expect optimality or approximation guarantees
---
## Numerical Experiment Design
See [references/numerical-experiments.md](references/numerical-experiments.md) for comprehensive guidance.
### Standard Experiment Structure
1. **Experimental Setup** (§X.1): Hardware, software, solver versions, random seeds
2. **Parameter Settings** (§X.2): Table of all parameters with default values and justification
3. **Benchmark Instances** (§X.3): Data source, generation method, instance sizes
4. **Main Results** (§X.4): Performance comparison tables with statistical tests
5. **Sensitivity Analysis** (§X.5): One-at-a-time or factorial design varying key parameters
6. **Managerial Implications of Experiments** (§X.6): What the numbers mean for practice
### MS&E-Specific Experiment Requirements
| Requirement | MS/OR | MSOM/POM | EJOR/IJPE | C&IE/IJPR |
|-------------|-------|----------|-----------|-----------|
| Real data instances | Preferred | Required | Strongly preferred | Optional |
| Statistical tests | Yes | Yes | Preferred | Optional |
| Sensitivity analysis | Required | Required | Required | Preferred |
| CPU time reporting | Required | Optional | Required | Required |
| Benchmark comparison | Required | Required | Required | Required |
| Instance size variety | Required | Required | Required | Required |
| Parameter justification | Required | Required | Required | Optional |
---
## Managerial Insights
See [references/managerial-insights.md](references/managerial-insights.md) for comprehensive guidance.
This is the **most distinctive feature** of MS&E papers. Managerial Insights (MIs) translate mathematical results into actionable business recommendations.
### MI Structure
Each managerial insight should follow the **SAR format**:
- **Setting**: Remind the reader of the context
- **Analytical finding**: State the mathematical result in plain language
- **Recommendation**: What a manager should DO differently
### MI Checklist
- [ ] Each MI is traceable to a specific theorem, proposition, or numerical result
- [ ] MIs use **zero mathematical notation**
- [ ] MIs are numbered for clarity (Managerial Insight 1, 2, 3...)
- [ ] Counterintuitive findings are explicitly highlighted
- [ ] Boundary conditions ("this insight holds only when...") are stated
- [ ] At least one MI connects to a real company or industry context
**Journals requiring mandatory MI**: MS, MSOM, OMEGA (critical), POM, IJPE.
**Journals where MI is optional but valued**: OR, TS, DS, EJOR.
---
## Writing Patterns
### Pattern 1: The Motivating Example
Open the introduction with a concrete mini-case from industry. Cite a real company, real numbers, real dilemma. This grounds the abstract model in practice. **Required by MS, MSOM, OMEGA.**
### Pattern 2: The Gap Table
Use a structured table comparing your paper against 8-12 key references on dimensions like: modeling approach, demand model, solution method, numerical validation, managerial insights. Explicitly mark your contribution. **Standard for MS, OR, EJOR.**
### Pattern 3: The Assumption-Relaxation Ladder
Present the base model → analyze → relax one assumption → re-analyze → relax another → re-analyze. Demonstrates robustness of the core insight. **Standard for MS, OR.**
### Pattern 4: The SAR Managerial Insight
Each MI is a self-contained mini-section: Setting → Analytical Finding → Recommendation. See [Managerial Insights](#managerial-insights) above.
---
## Venue-Specific Page Budgets
| Journal | Typical Submission Length | References | Appendix | Abstract Word Limit |
|---------|--------------------------|------------|----------|---------------------|
| MS | 38 pages | Unlimited | Unlimited | 200 |
| OR | 35 pages | Unlimited | Unlimited | 200 |
| MSOM | 32 pages | Unlimited | Unlimited | 200 |
| POM | 35 pages | Unlimited | Unlimited | 200 |
| TS | 32 pages | Unlimited | Unlimited | 200 |
| TRB | 32 pages | Unlimited | Unlimited | 150 |
| DS | 30 pages | Unlimited | Unlimited | 150 |
| OMEGA | ~6000 words | Unlimited | Limited | 150 |
| TRE | 30 pages | Unlimited | Unlimited | 200 |
| EJOR | 30 pages | Unlimited | Unlimited | 200 |
| IJPE | 30 pages | Unlimited | Unlimited | 200 |
| IJPR | 25 pages | Unlimited | Limited | 200 |
| C&IE | 25 pages | Unlimited | Limited | 200 |
---
## Cross-Referencing with Other Skills
This skill works with:
- **scientific-writing**: General IMRAD principles, clarity, citation formatting
- **literature-review**: Systematic literature search for MS&E topics
- **ppw:polish**: English polishing for non-native speakers (common in MS&E)
- **matlab** / **sympy**: For deriving and verifying mathematical results
---
## Common Rejection Reasons in MS&E Journals
| Reason | Frequency | Prevention |
|--------|-----------|------------|
| Insufficient contribution over existing literature | Very High | Explicit gap table in introduction |
| Unrealistic or undefended assumptions | High | Assumption defense section + relaxation |
| Managerial insights too generic | High | Specific, numbered, traceable to results |
| Numerical experiments not reproducible | Medium | Provide code, data, parameter justification |
| Proof errors or incomplete derivation | Medium | Have a colleague verify the proof chain |
| Poor motivation — why should anyone care? | High | Real-world motivating example |
| Wrong journal fit | Medium | Use the journal selection guide above |
| Paper too long for contribution size | Medium | Respect page budgets; move to appendix |
---
## 0-to-Draft Pipeline: Five-Stage Workflow
This is the core workflow of this skill. Each stage builds on the previous. Stages 1-3 are sequential; Stages 4-5 can partially overlap. At each stage, ask the user for their current materials before proceeding.
---
### Stage 1: Topic Positioning & Journal Selection
**Goal**: Define the paper's contribution gap, select a target journal, and establish the writing plan.
**Input needed from user**: Research area, preliminary idea, any existing notes or literature.
**Agent actions**:
```
1.1 Understand the research idea
- Ask: What is the managerial problem? What methodology? Any preliminary results?
- Identify the core trade-off or mechanism the paper will study
- Classify: Analytical / Empirical / Computational / Hybrid
1.2 Literature gap analysis
- Search for 8-12 closest papers using research-lookup or paper-lookup
- Build a Gap Table draft: rows = key references, columns = dimensions (methodology, demand model, solution approach, etc.)
- Identify the empty cell that this paper fills
- Output: draft gap table with explicit differentiation
1.3 Journal selection
- Use the Journal Selection Quick Reference and Decision Flowchart above
- Match paper profile (methodology, scope, contribution type) to journal identity
- Consider: MS (broad impact) vs. OR (method depth) vs. MSOM (empirical) vs. EJOR (method-inclusive)
- Output: recommended journal + 1-2 fallback options with rationale
1.4 Contribution statement
- Formulate the one-sentence contribution: "We [action] [what] using [method] and find that [key insight]."
- Draft 3-5 numbered contributions
- Output: contribution list
1.5 Writing plan
- Outline the paper structure with estimated page allocation per section
- Identify which reference files to load at each stage
- Output: section-by-section writing plan
```
**Stage 1 exit criteria**: Target journal chosen, gap table drafted, contribution list written, writing plan ready.
**Reference to load**: `references/journal-characteristics.md` (for chosen journal)
---
### Stage 2: Model Building
**Goal**: Design the mathematical model — the backbone of the paper.
**Input needed from user**: Problem description, decision variables, constraints, objective.
**Agent actions**:
```
2.1 Problem formalization
- Translate the verbal problem into mathematical structure
- Define: decision-maker, timeline (if dynamic), information structure (if stochastic)
- Output: verbal model description + event sequence
2.2 Notation system design
- Establish three-tier notation: sets (calligraphic), parameters (lowercase), variables (uppercase/bold)
- Check for symbol conflicts across sections
- Output: draft notation table
2.3 Assumption framework
- Enumerate assumptions as numbered statements
- For each: state the assumption, justify it, note what happens if relaxed
- Identify which assumptions are standard vs. novel to this paper
- Output: numbered assumption list with defenses
2.4 Model formulation
- Write objective function
- Write constraint set
- Write compact formulation: min/max {obj | constraints}
- Check: all symbols defined, all constraints numbered, boundary conditions explicit
- Output: §3 Model Formulation draft
2.5 Extensions planning
- Plan assumption-relaxation ladder: Base Model → Extension 1 → Extension 2 → ...
- Decide which extensions go in body vs. appendix
- Output: extensions outline
```
**Stage 2 exit criteria**: Complete model section draft with notation table, assumption list, base formulation, and extensions plan.
**Reference to load**: `references/modeling-conventions.md`
---
### Stage 3: Derivation & Analysis
**Goal**: Derive structural properties, prove main results, establish the analytical contribution.
**Input needed from user**: Model formulation (from Stage 2), any preliminary derivations.
**Agent actions**:
```
3.1 Structural property identification
- What properties should the optimal solution have? (convexity, monotonicity, threshold structure, etc.)
- Design the proof hierarchy: Lemma → Proposition → Theorem → Corollary
- Output: proof plan (what to prove, in what order)
3.2 Proof drafting
- For each result: state the theorem/proposition → provide intuition → proof/proof sketch
- Name the proof technique in the first sentence of each proof
- Check: all assumptions referenced, all steps justified, edge cases handled
- Decide body vs. appendix placement based on proof length and centrality
- Output: §4 Analysis section draft
3.3 Closed-form solutions (if applicable)
- Derive closed-form expressions for optimal decisions
- Verify boundary behavior (what happens at parameter limits?)
- Output: closed-form results with interpretation
3.4 Comparative statics
- How does the optimal solution change with key parameters?
- Use implicit function theorem, monotone comparative statics, or numerical exploration
- Output: comparative statics results
3.5 Proof verification checklist
- Every theorem stated (even if proof in appendix)
- Every statement has proof or proof sketch somewhere
- All "Proof." claims are verified
- No circular reasoning
```
**Stage 3 exit criteria**: Complete proof chain with all theorems/propositions/lemmas stated and proved.
**Reference to load**: `references/proof-derivation-guide.md`
---
### Stage 4: Algorithm & Numerical Experiments
**Goal**: Design solution method (if needed), run numerical experiments, generate figures and tables.
**Input needed from user**: Analytical results (from Stage 3), computational resources, data sources.
**Agent actions**:
```
4.1 Solution method design
- If model has closed-form solution → skip to 4.2
- If exact method: design algorithm, state complexity, prove optimality
- If heuristic: design procedure, calibrate parameters, validate on small instances
- Output: pseudocode + complexity analysis OR heuristic description
4.2 Experimental design
- Design parameter table: all parameters, base values, ranges, sources
- Determine benchmark instances: real data, literature instances, or generated
- Plan instance sizes: small/medium/large categories
- Design sensitivity analysis: which parameters, what ranges
- Output: experimental plan document
4.3 Code implementation guidance
- Suggest implementation approach (language, solver, key libraries)
- Provide pseudocode-to-code mapping for complex algorithms
- Remind: set random seeds, record solver versions, save all outputs
4.4 Results presentation
- Design performance comparison tables
- Design sensitivity analysis figures (line plots, heatmaps)
- Compute: optimality gaps, CPU times, statistical significance
- Draft figure captions that are self-contained
- Output: §6 Numerical Experiments draft with tables and figures
4.5 Managerial implications of experiments
- Connect numerical results back to analytical findings
- Translate numerical patterns into actionable recommendations
- Draft SAR-format insights from the experimental results
```
**Stage 4 exit criteria**: Complete experimental section with parameter table, benchmark description, results tables, sensitivity analysis, and experiment-derived insights.
**References to load**: `references/numerical-experiments.md`, `references/algorithm-solving-guide.md`
---
### Stage 5: Writing & Assembly
**Goal**: Write the full paper from Introduction to Conclusion, assemble all sections, polish.
**Input needed from user**: All prior stage outputs, target journal selection.
**Agent actions**:
```
5.1 Introduction drafting
- P1: Motivating example (real company, real numbers) — see Writing Pattern 1
- P2: Generalization to broader industry context
- P3-P4: Literature gap → "Despite its importance, the literature has..."
- P5: Our approach (one paragraph, no math)
- P6: Contributions (numbered, 3-5 items)
- P7: Paper organization ("The remainder of this paper is organized as follows...")
- Output: §1 Introduction draft
5.2 Literature review
- Organize by research stream, not paper-by-paper
- Include the Gap Table from Stage 1
- End each stream with: what's been done + what's missing
- Conclude with: how this paper bridges the gaps
- Output: §2 Literature Review draft
5.3 Managerial insights (standalone section)
- Distill analytical and numerical results into 3-6 SAR-format insights
- Each insight: Setting → Analytical Finding → Recommendation
- Zero mathematical notation
- Numbered, traceable to specific results
- Counterintuitive findings highlighted
- Output: §7 Managerial Insights draft
5.4 Conclusion
- Summary (2-3 sentences): problem, approach, key result
- Limitations (honest, specific): "Our analysis assumes X; relaxing this is future work."
- Future research directions (2-4 concrete directions)
- Output: §8 Conclusion draft
5.5 Abstract
- Write LAST, after the full paper is drafted
- Follow 5-sentence formula: (1) what achieved, (2) why hard/important,
(3) how done, (4) evidence, (5) key number/insight
- Max 200 words for MS/OR/MSOM; 150 for OMEGA/DS
- Output: Abstract
5.6 Cross-section consistency check
- Contributions list ↔ body sections (1:1 mapping)
- Notation: consistent across all sections
- All figures/tables referenced in text
- All citations in References section
- Managerial insights traceable to theorems/experiments
- Abstract claims substantiated in body
5.7 Formatting
- Apply journal-specific page budget (see table above)
- Move long proofs to appendix; keep proof sketches in body
- Format tables with booktabs; format algorithms with algorithm2e/algorithmicx
- Verify all citations programmatically (no hallucinated references)
```
**Stage 5 exit criteria**: Complete first draft with all sections, cross-checked for consistency, within page budget.
**References to load**: `references/writing-patterns.md`, `references/managerial-insights.md`, `references/reviewer-expectations.md`
---
### Pipeline Interaction Protocol
When the user invokes this skill, the agent should:
1. **Determine the current stage**: Ask what the user has so far (idea only? model built? results in hand?)
2. **Start at the appropriate stage**: Don't force users back to Stage 1 if they're ready for Stage 4
3. **At each stage**: Load the relevant reference file, then execute the agent actions
4. **Stage gate**: Confirm with user before proceeding to the next stage
5. **Iterate within stage**: Draft → get feedback → revise until the stage exit criteria are met
### Quick-Start Prompts
Users can jump to any stage:
- "I have a research idea about [topic]. Help me position it and pick a journal." → Stage 1
- "I have my model formulated. Help me derive structural properties." → Stage 3
- "I have analytical results. Help me design numerical experiments." → Stage 4
- "I have all my results. Help me write the full paper for MS." → Stage 5
- "Take me through the full pipeline from scratch." → Stage 1 → 5
---
## Quick Checklist Before Submission
- [ ] Target journal fits the paper's methodology-contribution profile
- [ ] Model section: all notation defined, assumptions stated and defended, extensions present
- [ ] Proofs: complete in body or appendix, proof techniques stated, markers (∎) present
- [ ] Algorithms: pseudocode present, complexity stated, approximation ratio if applicable
- [ ] Numerical experiments: parameter table, benchmark description, sensitivity analysis, CPU times
- [ ] Managerial insights: numbered, math-free, traceable to results, actionable
- [ ] Introduction: motivating example, gap table, numbered contributions, road map
- [ ] Literature review: organized by stream, not paper-by-paper; differentiation explicit
- [ ] Page budget: within journal limits (body only)
- [ ] All citations verified via DOI/programmatic fetch (no hallucinated references)
---
## References
- [references/journal-characteristics.md](references/journal-characteristics.md): Detailed per-journal characteristics, reviewer expectations, departmental editorial statements
- [references/modeling-conventions.md](references/modeling-conventions.md): Notation systems, assumption frameworks, model types, model section structure
- [references/proof-derivation-guide.md](references/proof-derivation-guide.md): Proof hierarchy, journal-specific proof standards, common proof techniques
- [references/numerical-experiments.md](references/numerical-experiments.md): Experiment design, parameter calibration, sensitivity analysis, benchmark instances
- [references/managerial-insights.md](references/managerial-insights.md): SAR framework, MI writing patterns, journal-specific MI expectations
- [references/reviewer-expectations.md](references/reviewer-expectations.md): What reviewers look for by journal, common criticisms, rebuttal strategies
- [references/writing-patterns.md](references/writing-patterns.md): Reusable writing patterns from published MS&E papers
© franklee16 · MIT · franklee16/academic-research-skills