name: literature-synthesis description: Systematic literature review and synthesis for Deep Research SOP Pipeline A. Use when starting research projects, conducting SOTA analysis, identifying research gaps, or preparing academic papers. Implements PRISMA-compliant systematic review methodology with automated search, screening, and synthesis across ArXiv, Semantic Scholar, and Papers with Code. version: 1.0.0 category: research tags:

• research
• analysis
• planning author: ruv

---

Literature Synthesis

Conduct systematic literature reviews following PRISMA guidelines, synthesizing state-of-the-art research to identify gaps and opportunities for Deep Research SOP Phase 1.

Overview

Purpose: Systematic literature review identifying SOTA methods, research gaps, and opportunities

When to Use:

• Starting new research projects (Phase 1 of Deep Research SOP)
• Conducting state-of-the-art (SOTA) analysis
• Identifying research gaps and opportunities
• Preparing related work sections for papers
• Validating novelty claims for proposed methods
• Quality Gate 1 requirement

Quality Gate: Required for Quality Gate 1 (minimum 50 papers)

Prerequisites:

• Research question formulated
• Search databases accessible (ArXiv, Semantic Scholar, Papers with Code)
• Reference management tool available (Zotero, Mendeley, BibTeX)

Outputs:

• Literature review document (50-100+ papers)
• SOTA performance benchmarks table
• Research gap analysis
• Hypothesis formulation
• Citation database (BibTeX)
• PRISMA flow diagram (if systematic review)

Time Estimate: 1-2 weeks

• Database search: 2-4 hours
• Screening: 1-2 days
• Full-text review: 3-5 days
• Synthesis: 2-3 days
• Writing: 1-2 days

Agents Used: researcher

---

Quick Start

1. Define Search Query

# Store research question in memory
npx claude-flow@alpha memory store \
  --key "sop/literature-review/research-question" \
  --value "How does multi-scale attention improve long-range dependency modeling in vision transformers?"

# Define search terms
search_terms="(multi-scale OR hierarchical) AND (attention OR transformer) AND (vision OR image)"

2. Database Search

# Search ArXiv
python scripts/search_arxiv.py \
  --query "$search_terms" \
  --start-date "2020-01-01" \
  --max-results 500 \
  --output literature/arxiv_results.json

# Search Semantic Scholar
python scripts/search_semantic_scholar.py \
  --query "$search_terms" \
  --fields "title,abstract,authors,year,citationCount,venue" \
  --min-citations 10 \
  --output literature/semantic_scholar_results.json

# Search Papers with Code
python scripts/search_papers_with_code.py \
  --task "image-classification" \
  --method "transformer" \
  --output literature/pwc_results.json

3. Screening and Selection

# Title/abstract screening
python scripts/screen_papers.py \
  --input literature/*_results.json \
  --inclusion-criteria literature/inclusion_criteria.yaml \
  --output literature/screened_papers.json

# Full-text review
python scripts/full_text_review.py \
  --input literature/screened_papers.json \
  --download-dir literature/pdfs/ \
  --output literature/selected_papers.json

4. Synthesis

# Extract SOTA benchmarks
python scripts/extract_sota_benchmarks.py \
  --papers literature/selected_papers.json \
  --datasets "ImageNet,CIFAR-10,CIFAR-100" \
  --output literature/sota_benchmarks.csv

# Identify research gaps
python scripts/identify_gaps.py \
  --papers literature/selected_papers.json \
  --output literature/research_gaps.md

5. Generate Literature Review

# Generate review document
python scripts/generate_literature_review.py \
  --papers literature/selected_papers.json \
  --benchmarks literature/sota_benchmarks.csv \
  --gaps literature/research_gaps.md \
  --template templates/literature_review_template.md \
  --output docs/literature_review.md

---

Detailed Instructions

Phase 1: Search Strategy Development (2-4 hours)

Objective: Define comprehensive search strategy

Steps:

1.1 Formulate Research Question (PICO Framework)

## Research Question (PICO)

**Population**: Vision transformer models
**Intervention**: Multi-scale attention mechanisms
**Comparison**: Standard single-scale attention
**Outcome**: Image classification accuracy, computational efficiency

1.2 Define Inclusion/Exclusion Criteria

# literature/inclusion_criteria.yaml

inclusion:
  - Published 2020 or later
  - Peer-reviewed (conferences: NeurIPS, ICML, CVPR, ICCV, ECCV; journals: TPAMI, IJCV)
  - Focuses on attention mechanisms in vision
  - Reports quantitative results on standard benchmarks
  - Available in English

exclusion:
  - Preprints without peer review (unless highly cited >100)
  - Not related to computer vision
  - No experimental results
  - Duplicate publications (keep most recent)
  - Surveys and position papers (catalog separately)

1.3 Define Search Terms

# scripts/search_terms.py

search_terms = {
    "attention_terms": [
        "attention mechanism",
        "self-attention",
        "multi-head attention",
        "cross-attention",
        "multi-scale attention"
    ],
    "architecture_terms": [
        "vision transformer",
        "ViT",
        "Swin Transformer",
        "hierarchical transformer",
        "pyramid transformer"
    ],
    "task_terms": [
        "image classification",
        "object detection",
        "semantic segmentation"
    ]
}

# Construct boolean query
query = " OR ".join(search_terms["attention_terms"]) + " AND " + \
        " OR ".join(search_terms["architecture_terms"])

Deliverable: Search strategy document

---

Phase 2: Database Search (2-4 hours)

Objective: Retrieve papers from multiple databases

Steps:

2.1 ArXiv Search

# scripts/search_arxiv.py
import arxiv

client = arxiv.Client()
search = arxiv.Search(
    query="(multi-scale OR hierarchical) AND (attention OR transformer) AND vision",
    max_results=500,
    sort_by=arxiv.SortCriterion.SubmittedDate,
    sort_order=arxiv.SortOrder.Descending
)

results = []
for paper in client.results(search):
    results.append({
        "title": paper.title,
        "authors": [author.name for author in paper.authors],
        "abstract": paper.summary,
        "published": paper.published.isoformat(),
        "arxiv_id": paper.entry_id.split("/")[-1],
        "pdf_url": paper.pdf_url,
        "categories": paper.categories
    })

# Save results
import json
with open("literature/arxiv_results.json", "w") as f:
    json.dump(results, f, indent=2)

print(f"Retrieved {len(results)} papers from ArXiv")

2.2 Semantic Scholar Search

# scripts/search_semantic_scholar.py
import requests

API_KEY = "YOUR_S2_API_KEY"
ENDPOINT = "https://api.semanticscholar.org/graph/v1/paper/search"

params = {
    "query": "(multi-scale OR hierarchical) AND (attention OR transformer) AND vision",
    "fields": "title,abstract,authors,year,citationCount,venue,publicationDate",
    "limit": 100,
    "minCitationCount": 10,
    "year": "2020-"
}

response = requests.get(ENDPOINT, params=params, headers={"x-api-key": API_KEY})
papers = response.json()["data"]

# Save results
with open("literature/semantic_scholar_results.json", "w") as f:
    json.dump(papers, f, indent=2)

print(f"Retrieved {len(papers)} papers from Semantic Scholar")

2.3 Papers with Code Search

# Download Papers with Code dataset
curl -o literature/pwc_dataset.json \
  https://paperswithcode.com/api/v1/papers/?tasks=image-classification&methods=transformer

# Filter relevant papers
python scripts/filter_pwc_papers.py \
  --input literature/pwc_dataset.json \
  --keywords "attention,multi-scale,hierarchical" \
  --output literature/pwc_results.json

Deliverable: Combined database of ~200-500 papers

---

Phase 3: Screening (1-2 days)

Objective: Screen papers based on inclusion/exclusion criteria

Steps:

3.1 Remove Duplicates

# scripts/remove_duplicates.py
from fuzzywuzzy import fuzz

def find_duplicates(papers, threshold=90):
    duplicates = []
    for i, p1 in enumerate(papers):
        for j, p2 in enumerate(papers[i+1:], i+1):
            similarity = fuzz.ratio(p1["title"], p2["title"])
            if similarity >= threshold:
                # Keep the one with more citations
                keep = i if p1.get("citationCount", 0) >= p2.get("citationCount", 0) else j
                remove = j if keep == i else i
                duplicates.append(remove)
    return list(set(duplicates))

# Remove duplicates
papers = load_papers("literature/*_results.json")
duplicates = find_duplicates(papers)
unique_papers = [p for i, p in enumerate(papers) if i not in duplicates]

print(f"Removed {len(duplicates)} duplicates, {len(unique_papers)} unique papers remain")

3.2 Title/Abstract Screening

# scripts/screen_papers.py
import yaml

# Load inclusion criteria
with open("literature/inclusion_criteria.yaml") as f:
    criteria = yaml.safe_load(f)

def screen_paper(paper, criteria):
    """Screen paper based on title and abstract."""
    # Check publication year
    year = int(paper.get("year", paper.get("published", "2000")[:4]))
    if year < 2020:
        return False, "Published before 2020"

    # Check for relevant keywords in title/abstract
    text = f"{paper['title']} {paper.get('abstract', '')}".lower()
    relevant_keywords = ["attention", "transformer", "vision", "image"]
    if not any(kw in text for kw in relevant_keywords):
        return False, "Not relevant to research question"

    # Check for exclusion criteria
    if "survey" in text or "review" in text:
        return False, "Survey or review paper"

    return True, "Included"

# Screen all papers
screened = []
for paper in unique_papers:
    include, reason = screen_paper(paper, criteria)
    paper["screening_decision"] = "include" if include else "exclude"
    paper["screening_reason"] = reason
    if include:
        screened.append(paper)

print(f"{len(screened)}/{len(unique_papers)} papers passed title/abstract screening")

Deliverable: ~50-100 papers after screening

---

Phase 4: Full-Text Review (3-5 days)

Objective: Review full papers for final inclusion

Steps:

4.1 Download PDFs

# scripts/download_pdfs.py
import requests
from pathlib import Path

pdf_dir = Path("literature/pdfs")
pdf_dir.mkdir(exist_ok=True)

for paper in screened_papers:
    pdf_url = paper.get("pdf_url") or paper.get("openAccessPdf", {}).get("url")
    if pdf_url:
        filename = f"{paper['arxiv_id'] or paper['paperId']}.pdf"
        response = requests.get(pdf_url)
        (pdf_dir / filename).write_bytes(response.content)
        print(f"Downloaded: {filename}")

4.2 Extract Key Information

# scripts/extract_paper_info.py

def extract_info(paper_pdf):
    """Extract key information from paper."""
    info = {
        "methods": [],  # Novel methods proposed
        "datasets": [],  # Datasets used
        "metrics": {},  # Performance metrics
        "comparisons": [],  # Baseline comparisons
        "limitations": [],  # Reported limitations
        "future_work": []  # Suggested future work
    }

    # Use PyPDF2 or pdfplumber to extract text
    # Use regex or NLP to extract structured information
    # (Implementation details omitted for brevity)

    return info

# Extract info from all papers
for paper in screened_papers:
    pdf_path = f"literature/pdfs/{paper['arxiv_id']}.pdf"
    if Path(pdf_path).exists():
        paper["extracted_info"] = extract_info(pdf_path)

4.3 Final Inclusion Decision

# Manual review criteria
def final_review(paper):
    """Final inclusion decision based on full text."""
    info = paper.get("extracted_info", {})

    # Must report quantitative results
    if not info.get("metrics"):
        return False, "No quantitative results"

    # Must use standard benchmarks
    standard_datasets = ["ImageNet", "CIFAR-10", "CIFAR-100", "COCO"]
    if not any(ds in info.get("datasets", []) for ds in standard_datasets):
        return False, "No standard benchmarks"

    # Must compare with baselines
    if not info.get("comparisons"):
        return False, "No baseline comparisons"

    return True, "Included"

# Final review
selected_papers = []
for paper in screened_papers:
    include, reason = final_review(paper)
    if include:
        selected_papers.append(paper)

print(f"{len(selected_papers)} papers selected for synthesis")

Deliverable: 50-70 final selected papers

---

Phase 5: Synthesis (2-3 days)

Objective: Synthesize findings into coherent narrative

Steps:

5.1 Extract SOTA Benchmarks

# scripts/extract_sota_benchmarks.py
import pandas as pd

benchmarks = []
for paper in selected_papers:
    metrics = paper["extracted_info"]["metrics"]
    for dataset, results in metrics.items():
        benchmarks.append({
            "paper": paper["title"],
            "year": paper["year"],
            "method": paper["extracted_info"]["methods"][0] if paper["extracted_info"]["methods"] else "N/A",
            "dataset": dataset,
            "metric": results.get("accuracy") or results.get("top1"),
            "citation_count": paper.get("citationCount", 0)
        })

df = pd.DataFrame(benchmarks)
df.to_csv("literature/sota_benchmarks.csv", index=False)

# Find best performance per dataset
sota = df.groupby("dataset")["metric"].max()
print("State-of-the-Art Performance:")
print(sota)

5.2 Identify Research Gaps

# scripts/identify_gaps.py

gaps = {
    "methodological_gaps": [],
    "application_gaps": [],
    "evaluation_gaps": []
}

# Methodological gaps
methods_used = set()
for paper in selected_papers:
    methods_used.update(paper["extracted_info"]["methods"])

# Check for unexplored combinations
all_attention_types = ["self-attention", "cross-attention", "multi-scale", "sparse", "local", "global"]
explored = [m for m in all_attention_types if any(m in method for method in methods_used)]
unexplored = [m for m in all_attention_types if m not in explored]

gaps["methodological_gaps"] = unexplored

# Application gaps
datasets_used = set()
for paper in selected_papers:
    datasets_used.update(paper["extracted_info"]["datasets"])

standard_datasets = ["ImageNet", "CIFAR-10", "COCO", "ADE20K", "Cityscapes"]
underexplored = [d for d in standard_datasets if d not in datasets_used]

gaps["application_gaps"] = underexplored

# Write gaps analysis
with open("literature/research_gaps.md", "w") as f:
    f.write("# Research Gaps Analysis\n\n")
    f.write("## Methodological Gaps\n")
    for gap in gaps["methodological_gaps"]:
        f.write(f"- {gap}: Not explored in reviewed papers\n")
    f.write("\n## Application Gaps\n")
    for gap in gaps["application_gaps"]:
        f.write(f"- {gap}: Underexplored dataset\n")

5.3 Formulate Hypotheses

# Hypotheses (based on research gaps)

## H1: Multi-Scale Local-Global Attention
**Gap**: No papers combine local and global attention at multiple scales
**Hypothesis**: Combining local (window-based) and global (full-image) attention at multiple scales will improve long-range dependency modeling while maintaining computational efficiency
**Expected Improvement**: +2-5% accuracy on ImageNet
**Testable**: Compare multi-scale local-global vs. single-scale global attention

## H2: Sparse Attention for High-Resolution Images
**Gap**: Limited exploration of sparse attention for high-resolution inputs (>1024px)
**Hypothesis**: Learned sparse attention patterns can reduce computational cost O(n²) → O(n log n) for high-resolution images without accuracy loss
**Expected Improvement**: 2-3x speedup with <1% accuracy drop
**Testable**: Compare sparse vs. dense attention on high-res datasets

Deliverable: Research gaps and hypotheses document

---

Phase 6: Writing (1-2 days)

Objective: Write comprehensive literature review

Steps:

6.1 Structure Literature Review

# Literature Review: Multi-Scale Attention in Vision Transformers

## 1. Introduction
- Research question and motivation
- Scope of review (50+ papers, 2020-2025)
- Organization of review

## 2. Background
- Vision Transformers fundamentals
- Attention mechanisms overview
- Multi-scale representations in vision

## 3. Attention Mechanisms in Vision
### 3.1 Self-Attention
- Vanilla self-attention (Vaswani et al., 2017)
- Vision Transformer (Dosovitskiy et al., 2021)

### 3.2 Hierarchical Attention
- Swin Transformer (Liu et al., 2021)
- Pyramid Vision Transformer (Wang et al., 2021)

### 3.3 Multi-Scale Attention
- CrossViT (Chen et al., 2021)
- Multi-Scale Vision Transformer (Fan et al., 2021)

## 4. State-of-the-Art Performance
[Table of SOTA benchmarks]

## 5. Research Gaps
- Methodological gaps
- Application gaps
- Evaluation gaps

## 6. Proposed Direction
- Hypotheses based on gaps
- Expected contributions

## 7. Conclusion

6.2 Generate BibTeX Citations

# scripts/generate_bibtex.py

def paper_to_bibtex(paper):
    """Convert paper to BibTeX entry."""
    authors = " and ".join(paper["authors"])
    year = paper["year"]
    title = paper["title"]
    venue = paper.get("venue", "arXiv")

    bibtex_key = f"{paper['authors'][0].split()[-1]}{year}"

    return f"""@inproceedings{{{bibtex_key},
  title={{{title}}},
  author={{{authors}}},
  booktitle={{{venue}}},
  year={{{year}}}
}}"""

# Generate bibliography
with open("literature/bibliography.bib", "w") as f:
    for paper in selected_papers:
        f.write(paper_to_bibtex(paper) + "\n\n")

Deliverable: Complete literature review document

---

Phase 7: PRISMA Flow Diagram (Optional, 1 hour)

Objective: Create PRISMA flow diagram for systematic reviews

# scripts/generate_prisma_diagram.py

prisma_data = {
    "identification": {
        "records_identified": 500,
        "duplicates_removed": 150
    },
    "screening": {
        "records_screened": 350,
        "records_excluded": 250
    },
    "eligibility": {
        "full_text_assessed": 100,
        "full_text_excluded": 30,
        "exclusion_reasons": {
            "no_quantitative_results": 15,
            "no_standard_benchmarks": 10,
            "not_peer_reviewed": 5
        }
    },
    "included": {
        "studies_included": 70
    }
}

# Generate PRISMA diagram using GraphViz
# (Implementation omitted for brevity)

Deliverable: PRISMA flow diagram

---

Integration with Deep Research SOP

Pipeline Integration

• Pipeline A (Literature Synthesis): This skill implements the complete literature review phase
• Phase 1 (Foundations): Required before baseline replication
• Quality Gate 1: Minimum 50 papers required

Agent Coordination

researcher agent:
  - Conducts database searches
  - Screens papers
  - Extracts key information
  - Synthesizes findings
  - Formulates hypotheses

Memory Coordination

# Store literature review in memory
npx claude-flow@alpha memory store \
  --key "sop/phase1/literature-review" \
  --value "$(cat docs/literature_review.md)"

# Store SOTA benchmarks
npx claude-flow@alpha memory store \
  --key "sop/phase1/sota-benchmarks" \
  --value "$(cat literature/sota_benchmarks.csv)"

---

Troubleshooting

Issue: Too few papers found (<50)

Solution: Broaden search terms, expand date range, include preprints

Issue: Too many papers (>200)

Solution: Add inclusion criteria (minimum citations, specific venues)

Issue: Gate 1 validation fails due to incomplete review

Solution: Ensure ≥50 papers reviewed, SOTA benchmarks documented, research gaps identified

---

Related Skills and Commands

Prerequisites

• None (first skill in Deep Research SOP workflow)

Next Steps

• baseline-replication - Replicate SOTA baselines identified in literature review
• gate-validation --gate 1 - Validate Phase 1 completion

Related Commands

• /prisma-init - Initialize PRISMA systematic review (researcher agent)

---

References

Systematic Review Methodologies

• PRISMA 2020 (Page et al., 2021): Systematic review reporting guidelines
• Cochrane Handbook: Gold standard for systematic reviews

Search Databases

• ArXiv: https://arxiv.org/
• Semantic Scholar: https://www.semanticscholar.org/
• Papers with Code: https://paperswithcode.com/

Reference Management

• Zotero: https://www.zotero.org/
• BibTeX format specification

---

Core Principles

Literature Synthesis operates on 3 fundamental principles:

Principle 1: PRISMA-Compliant Systematic Methodology

All literature reviews follow PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines with documented search strategy, inclusion/exclusion criteria, and screening process. This ensures reproducible and comprehensive coverage.

In practice:

• Pre-registered search terms and databases before screening begins
• PRISMA flow diagram documents identification, screening, eligibility, and inclusion phases
• Minimum 50 papers reviewed for Quality Gate 1 compliance

Principle 2: Multi-Source Triangulation for SOTA Validation

State-of-the-art claims are cross-validated across ArXiv, Semantic Scholar, and Papers with Code to identify consensus performance benchmarks and conflicting results. This prevents reliance on single-source biases.

In practice:

• SOTA benchmarks extracted from 3+ independent sources (papers, leaderboards, reproducibility studies)
• Conflicting results reported with methodology analysis to explain divergence
• Citation counts and venue tier (h-index, acceptance rates) used as credibility filters

Principle 3: Transparent Research Gap Identification

Research gaps are identified through systematic evidence rather than intuition, with explicit documentation of what was searched, what was found, and what remains unexplored. This enables hypothesis formulation grounded in literature.

In practice:

• Methodological gaps: Attention types explored vs unexplored (self, cross, multi-scale, sparse)
• Application gaps: Datasets used vs underexplored (ImageNet covered, Cityscapes underexplored)
• Evaluation gaps: Metrics reported (accuracy) vs missing (fairness, robustness)

Common Anti-Patterns

Anti-Pattern	Problem	Solution
Cherry-Picking Papers	Selecting only papers supporting predetermined conclusions creates confirmation bias and misses contradictory evidence	Use systematic search with pre-registered criteria, report contradictory findings explicitly
Claiming Gaps Without Evidence	Asserting novelty without systematic search allows claims like "no prior work on X" when X exists in literature	Document search queries used, databases consulted, and negative results (searched but found nothing)
Single-Source SOTA Benchmarks	Relying on one paper's reported SOTA creates vulnerability to errors, non-reproducible results, or outdated information	Cross-validate SOTA across Papers with Code leaderboards, reproducibility studies, and multiple independent papers

Conclusion

Literature Synthesis provides PRISMA-compliant systematic literature review methodology for identifying state-of-the-art methods, performance benchmarks, and research gaps. By enforcing multi-source triangulation, transparent search documentation, and explicit gap analysis, this skill ensures literature reviews meet academic publication standards and Quality Gate 1 requirements.

Use this skill at the start of research projects (Deep Research SOP Phase 1) when conducting SOTA analysis, preparing related work sections, or validating novelty claims. The 7-phase workflow (search strategy, database search, screening, full-text review, synthesis, writing, PRISMA documentation) produces comprehensive reviews with 50-100+ papers, SOTA benchmark tables, and evidence-backed research gap identification. The result is a solid foundation for hypothesis formulation and method development with clear understanding of what exists, what performs best, and what remains unexplored.

---