Citation Chasing & Mapping

Explore and map citation networks to discover related research, identify foundational papers, and trace the evolution of scientific fields. Supports both backward citation chasing (finding cited works/ancestors) and forward citation chasing (finding citing works/descendants) for comprehensive literature reviews.

Note: This tool exports citation network data in JSON format for use with external visualization tools (e.g., Cytoscape, Gephi). It does not generate visualizations directly.

Key Capabilities:

Bidirectional Citation Tracing: Find both papers cited by a work and papers that cite it
Multi-Hop Network Analysis: Explore citation networks up to 3 degrees of separation
Foundational Paper Identification: Discover highly-cited works that shaped the field
Field Evolution Mapping: Track how research topics developed over time

When to Use

✅ Use this skill when:

Conducting comprehensive literature reviews and need to ensure completeness
Starting research in a new field and need to identify key papers
Writing a review article and need to map the development of a topic
Identifying foundational works that everyone in the field should know
Finding recent developments that cite a seminal paper
Tracing the evolution of an idea from its origin to present day
Discovering related research that might not appear in keyword searches
Preparing grant proposals and need to establish the intellectual foundation

❌ Do NOT use when:

Looking for full-text papers → Use literature-full-text-fetcher or pubmed-central-downloader
Needing bibliographic management → Use Zotero, Mendeley, or EndNote
Performing systematic reviews with strict inclusion criteria → Use Covidence or similar tools
Requiring citation counts and metrics only → Use Google Scholar, Web of Science
Analyzing co-authorship networks → Use specialized social network tools
Doing topic modeling or text mining → Use NLP tools like LDA or BERT
Needing real-time citation alerts → Use journal alert services or RSS feeds

Related Skills:

上游 (Upstream): literature-full-text-fetcher, abstract-summarizer
下游 (Downstream): journal-club-presenter, cross-disciplinary-bridge-finder

Integration with Other Skills

Upstream Skills:

literature-full-text-fetcher: Retrieve full text of papers identified through citation chasing
abstract-summarizer: Summarize papers found in citation network
conference-abstract-adaptor: Find related conference presentations

Downstream Skills:

journal-club-presenter: Create presentations about key papers identified
cross-disciplinary-bridge-finder: Find connections to other fields
grant-proposal-assistant: Use identified papers for background sections

Complete Workflow:

Seed Paper → citation-chasing-mapping → literature-full-text-fetcher → abstract-summarizer → Comprehensive Review

Core Capabilities

1. Backward Citation Chasing (Ancestors)

Find papers cited by a target paper to trace intellectual foundations and prior work.

from scripts.main import CitationNetwork

# Initialize network
network = CitationNetwork()

# Add papers with their citations
network.add_paper(
    "CRISPR_2012",
    "A Programmable Dual-RNA-Guided DNA Endonuclease", 
    2012,
    citations=["ZFN_1996", "TALEN_2011", "tracrRNA_2008", "Cas9_2005"]
)

network.add_paper(
    "ZFN_1996",
    "Zinc Finger Nucleases",
    1996,
    citations=["zinc_finger_1985"]
)

# Find papers cited by CRISPR 2012 (depth 1)
seed_paper = "CRISPR_2012"
depth = 1

if seed_paper in network.papers:
    direct_citations = network.papers[seed_paper]["citations"]
    print(f"Papers cited by '{network.papers[seed_paper]['title']}' (depth 1):")
    for cited_id in direct_citations:
        if cited_id in network.papers:
            cited = network.papers[cited_id]
            print(f"  [{cited_id}] {cited['title']} ({cited['year']})")

Backward Chasing Strategy:

Depth	What You Find	Use Case
1 hop	Direct citations	Immediate intellectual foundation
2 hops	Citations of citations	Broader field context
3 hops	Extended network	Historical development

Best Practices:

✅ Start with recent review papers - they cite comprehensively
✅ Note citation patterns - papers cited multiple times are foundational
✅ Check publication years - ensure chronological logic
✅ Look for review papers in the citation list for broader context

Common Issues and Solutions:

Issue: Too many citations to review

Symptom: High-impact papers have hundreds of citations
Solution: Focus on papers cited multiple times; prioritize reviews

Issue: Circular references

Symptom: Papers cite each other (A cites B, B cites A)
Solution: Check publication dates to establish true lineage

2. Forward Citation Chasing (Descendants)

Find papers that cite a target paper to discover recent developments and applications.

from scripts.main import CitationNetwork

network = CitationNetwork()

# Add papers
network.add_paper("CRISPR_2012", "CRISPR-Cas9 Original", 2012, [])
network.add_paper("CRISPR_App_2013", "CRISPR Applications", 2013, ["CRISPR_2012"])
network.add_paper("CRISPR_Therapy_2015", "Therapeutic CRISPR", 2015, ["CRISPR_2012", "CRISPR_App_2013"])
network.add_paper("CRISPR_BaseEdit_2016", "Base Editing", 2016, ["CRISPR_2012"])
network.add_paper("CRISPR_Clinical_2018", "Clinical Trials", 2018, ["CRISPR_Therapy_2015"])

# Find papers that cite CRISPR_2012 (forward chasing)
seed_paper = "CRISPR_2012"
citing_papers = network.find_citing_papers(seed_paper)

print(f"Papers citing '{network.papers[seed_paper]['title']}':")
for citing_id in citing_papers:
    citing = network.papers[citing_id]
    print(f"  [{citing_id}] {citing['title']} ({citing['year']})")

# Multi-hop forward search
related = network.find_related_papers(seed_paper, depth=2)
print(f"\nExtended network (2 hops): {len(related)} papers")

Forward Chasing Strategy:

Approach	Best For	Expected Results
Direct citations	Immediate follow-up work	50-500 papers
2-hop network	Field development	500-2000 papers
Filtered by year	Recent advances	Most recent subset
Filtered by topic	Specific applications	Focused subset

Best Practices:

✅ Check citation context - not all citations are substantive
✅ Sort by year - identify recent developments
✅ Look for application papers - show practical impact
✅ Track methodological improvements - technical advances

Common Issues and Solutions:

Issue: Too many citing papers

Symptom: Seminal papers have thousands of citations
Solution: Filter by year, journal impact, or citation context

Issue: Citation without relevance

Symptom: Paper cited only in introduction/background
Solution: Read abstracts to assess substantive engagement

3. Foundational Paper Identification

Identify highly-cited papers that represent turning points or foundational work in a field.

from scripts.main import CitationNetwork

# Build network from file or demo
network = CitationNetwork()

# Add sample papers
papers_data = [
    ("Genomics_2001", "Human Genome Paper", 2001, []),
    ("NGS_2005", "Next-Gen Sequencing", 2005, ["Genomics_2001"]),
    ("RNA_Seq_2008", "RNA-Seq Methods", 2008, ["NGS_2005"]),
    ("scRNA_2015", "Single-Cell RNA-Seq", 2015, ["RNA_Seq_2008", "NGS_2005"]),
    ("ATAC_2013", "ATAC-Seq", 2013, ["NGS_2005"]),
]

for pid, title, year, citations in papers_data:
    network.add_paper(pid, title, year, citations)

# Identify most cited papers
key_papers = network.identify_key_papers()

print("\n🏆 Foundational Papers (Most Cited):")
print("-" * 60)
for rank, (pid, count) in enumerate(key_papers[:5], 1):
    paper = network.papers.get(pid, {})
    title = paper.get('title', 'Unknown')
    year = paper.get('year', '?')
    print(f"{rank}. [{pid}] ({year})")
    print(f"   Citations: {count}")
    print(f"   Title: {title}")
    print()

Foundational Paper Indicators:

Indicator	Significance	Example
High citation count	Broad impact	>1000 citations
Cited by diverse fields	Cross-disciplinary	Multiple domain papers
Early publication	Foundational	Early in field timeline
Methodology paper	Enables research	New technique/protocol
Review in top journal	Synthesis	Nature/Science/Cell reviews

Best Practices:

✅ Consider citation velocity - recent papers with rapid uptake
✅ Check citation context - meaningful vs. perfunctory citations
✅ Look for methodological papers - often highly cited
✅ Identify review papers - comprehensive field overviews

Common Issues and Solutions:

Issue: Methodology vs. Discovery papers

Symptom: Methods papers often have more citations than discovery papers
Solution: Separate analysis by paper type; both are valuable

Issue: Self-citations inflate counts

Symptom: Authors cite their own work extensively
Solution: Check for self-citation patterns; focus on external citations

4. Multi-Hop Network Analysis

Explore citation networks up to 3 degrees of separation for comprehensive field mapping.

from scripts.main import CitationNetwork

network = CitationNetwork()

# Build a more complex network
papers = {
    "A_2000": {"title": "Original Discovery", "year": 2000, "citations": []},
    "B_2002": {"title": "First Application", "year": 2002, "citations": ["A_2000"]},
    "C_2003": {"title": "Method Improvement", "year": 2003, "citations": ["A_2000"]},
    "D_2005": {"title": "Clinical Study", "year": 2005, "citations": ["B_2002"]},
    "E_2006": {"title": "Large Scale Study", "year": 2006, "citations": ["C_2003"]},
    "F_2008": {"title": "Meta-Analysis", "year": 2008, "citations": ["D_2005", "E_2006"]},
}

for pid, data in papers.items():
    network.add_paper(pid, data["title"], data["year"], data["citations"])

# Analyze network from seed paper
seed = "A_2000"

for depth in [1, 2, 3]:
    related = network.find_related_papers(seed, depth=depth)
    print(f"\nDepth {depth}: {len(related)} related papers")
    
    # Group by generation
    if depth == 1:
        print("  Generation 1 (direct citations):")
    elif depth == 2:
        print("  Generation 2 (citations of citations):")
    else:
        print("  Generation 3 (extended network):")
    
    for pid in sorted(related)[:5]:  # Show first 5
        if pid in network.papers:
            paper = network.papers[pid]
            print(f"    [{pid}] {paper['title']} ({paper['year']})")
    
    if len(related) > 5:
        print(f"    ... and {len(related) - 5} more")

Network Depth Analysis:

Depth	Coverage	Use Case	Risk
1 hop	Direct neighbors	Focused review	May miss context
2 hops	Extended network	Comprehensive	Manageable size
3 hops	Broad field	Historical mapping	Too large/diluted

Best Practices:

✅ Start with depth=2 for most reviews
✅ Check diminishing returns - additional hops add noise
✅ Prune by relevance - remove off-topic papers
✅ Map chronologically - understand temporal development

Common Issues and Solutions:

Issue: Network explosion

Symptom: 3 hops generates thousands of papers
Solution: Use depth=2 with filtering; or sample from 3-hop results

Issue: Off-topic drift

Symptom: Distant papers lose connection to original topic
Solution: Apply topic filters; manually curate distant papers

5. Network Data Export

Export citation network data for external analysis and visualization tools.

from scripts.main import CitationNetwork
import json

network = CitationNetwork()

# Add sample network
network.add_paper("Paper_A", "Original Work", 2010, ["Paper_B", "Paper_C"])
network.add_paper("Paper_B", "Prior Method", 2008, ["Paper_D"])
network.add_paper("Paper_C", "Related Theory", 2009, ["Paper_D"])
network.add_paper("Paper_D", "Foundational Work", 2005, [])
network.add_paper("Paper_E", "Follow-up Study", 2012, ["Paper_A"])

output_file = "citation_network.json"
network.export_network(output_file)

# Preview structure
with open(output_file, 'r') as f:
    data = json.load(f)

print("Network Structure:")
print(f"  Nodes (papers): {len(data['nodes'])}")
print(f"  Edges (citations): {len(data['edges'])}")

print("\nNode example:")
print(json.dumps(data['nodes'][0], indent=2))

print("\nEdge example:")
print(json.dumps(data['edges'][0], indent=2))

# External visualization tools
print("\n" + "="*60)
print("RECOMMENDED EXTERNAL TOOLS FOR VISUALIZATION:")
print("="*60)
print("1. Cytoscape: Import JSON, use 'yFiles Circular Layout'")
print("2. Gephi: Import JSON, use 'Force Atlas 2' layout")
print("3. D3.js: Use force-directed graph example")
print("4. Python: networkx + matplotlib (separate installation required)")

Network Format Specification:

{
  "nodes": [
    {
      "id": "paper_id",
      "title": "Paper Title",
      "year": 2020,
      "citation_count": 45
    }
  ],
  "edges": [
    {
      "source": "citing_paper",
      "target": "cited_paper"
    }
  ]
}

Note: This tool exports network data in standard JSON format for use with external visualization tools. It does not generate visualizations directly.

6. Field Evolution Mapping

Trace how a research field developed over time by analyzing citation patterns chronologically.

from scripts.main import CitationNetwork
from collections import defaultdict

network = CitationNetwork()

# Add papers across multiple years
field_papers = [
    ("Genomics_2001", "Human Genome", 2001, []),
    ("Microarray_2002", "Expression Arrays", 2002, ["Genomics_2001"]),
    ("RNAi_2003", "RNA Interference", 2003, []),
    ("NGS_2005", "Next-Gen Seq", 2005, ["Genomics_2001"]),
    ("RNA_Seq_2008", "RNA-Seq", 2008, ["NGS_2005", "Microarray_2002"]),
    ("GWAS_2009", "GWAS Methods", 2009, ["Genomics_2001"]),
    ("scRNA_2015", "Single-Cell", 2015, ["RNA_Seq_2008", "NGS_2005"]),
    ("CRISPR_2012", "CRISPR", 2012, ["RNAi_2003"]),
    ("Spatial_2020", "Spatial Omics", 2020, ["scRNA_2015", "RNA_Seq_2008"]),
]

for pid, title, year, citations in field_papers:
    network.add_paper(pid, title, year, citations)

# Analyze evolution by year
papers_by_year = defaultdict(list)
for pid, data in network.papers.items():
    papers_by_year[data['year']].append(pid)

print("\n📊 Field Evolution Timeline:")
print("="*60)

for year in sorted(papers_by_year.keys()):
    papers = papers_by_year[year]
    print(f"\n{year}:")
    for pid in papers:
        paper = network.papers[pid]
        citations = paper['citations']
        if citations:
            cited_titles = [network.papers.get(c, {}).get('title', '?')[:20] for c in citations]
            citation_str = f" ← {', '.join(cited_titles)}"
        else:
            citation_str = " (foundational)"
        print(f"  • {paper['title'][:40]}{citation_str}")

# Identify field shifts
print("\n" + "="*60)
print("🔍 Field Development Insights:")
print("="*60)

# Find periods of high activity
for year in sorted(papers_by_year.keys()):
    count = len(papers_by_year[year])
    if count >= 2:
        print(f"{year}: Active year with {count} major papers")

Evolution Analysis Metrics:

Metric	What It Shows	Interpretation
Publication rate	Field activity	Growth or maturity
Citation patterns	Knowledge flow	Influential works
Method transitions	Technique changes	Technical progress
Cross-citation	Subfield integration	Field convergence

Best Practices:

✅ Track methodological papers - often mark transitions
✅ Identify review paper timing - indicates field maturity
✅ Note citation half-life - rate of knowledge turnover
✅ Map paradigm shifts - major changes in approach

Common Issues and Solutions:

Issue: Confounding by publication delays

Symptom: Recent papers appear less influential
Solution: Normalize by years since publication; use citation velocity

Complete Workflow Example

From seed paper to comprehensive literature map:

# Step 1: Create sample citation network
python scripts/main.py --paper-id "CRISPR_2012" --depth 2 --output network.json

# Step 2: View key foundational papers
python scripts/main.py --network-file network.json

Python API Usage:

from scripts.main import CitationNetwork
import json

def conduct_comprehensive_review(
    seed_papers: list,
    max_depth: int = 2,
    output_dir: str = "./literature_review"
) -> dict:
    """
    Comprehensive citation-based literature review.
    
    Args:
        seed_papers: List of seed paper IDs to start from
        max_depth: Maximum citation hops
        output_dir: Directory for output files
        
    Returns:
        Dictionary with review statistics and findings
    """
    from pathlib import Path
    Path(output_dir).mkdir(parents=True, exist_ok=True)
    
    network = CitationNetwork()
    
    # Build network (in real use, load from database)
    # This example uses demo data
    demo_papers = {
        "Seed_2020": {"title": "Recent Discovery", "year": 2020, "citations": ["Foundational_2015", "Method_2018"]},
        "Foundational_2015": {"title": "Original Theory", "year": 2015, "citations": ["Classical_2010"]},
        "Method_2018": {"title": "Key Method", "year": 2018, "citations": ["Foundational_2015"]},
        "Classical_2010": {"title": "Seminal Work", "year": 2010, "citations": []},
        "FollowUp_2021": {"title": "Application Study", "year": 2021, "citations": ["Seed_2020"]},
        "Review_2022": {"title": "Recent Review", "year": 2022, "citations": ["Seed_2020", "Method_2018"]},
    }
    
    for pid, data in demo_papers.items():
        network.add_paper(pid, data["title"], data["year"], data.get("citations", []))
    
    # Analyze each seed paper
    all_related = set()
    findings = {
        "seed_papers": [],
        "foundational_papers": [],
        "recent_developments": [],
        "network_stats": {}
    }
    
    print("\n" + "="*70)
    print("COMPREHENSIVE LITERATURE REVIEW")
    print("="*70)
    
    for seed in seed_papers:
        if seed not in network.papers:
            print(f"⚠️  Seed paper '{seed}' not found in database")
            continue
            
        print(f"\n🎯 Analyzing: {seed}")
        print(f"   Title: {network.papers[seed]['title']}")
        
        # Find related papers
        related = network.find_related_papers(seed, max_depth)
        all_related.update(related)
        
        # Categorize findings
        for pid in related:
            if pid not in network.papers:
                continue
            paper = network.papers[pid]
            
            if paper['year'] <= 2015:
                findings["foundational_papers"].append(pid)
            elif paper['year'] >= 2020:
                findings["recent_developments"].append(pid)
        
        print(f"   Found {len(related)} related papers")
    
    # Identify key papers across entire network
    key_papers = network.identify_key_papers()
    
    # Export network
    network_file = f"{output_dir}/citation_network.json"
    network.export_network(network_file)
    
    # Compile summary
    summary = {
        "total_papers_analyzed": len(network.papers),
        "seed_papers": len(seed_papers),
        "related_papers_found": len(all_related),
        "foundational_papers": len(findings["foundational_papers"]),
        "recent_papers": len(findings["recent_developments"]),
        "top_cited_papers": [(pid, count) for pid, count in key_papers[:5]],
        "output_files": {
            "network_json": network_file
        }
    }
    
    # Print summary
    print("\n" + "="*70)
    print("REVIEW SUMMARY")
    print("="*70)
    print(f"Total papers in database: {summary['total_papers_analyzed']}")
    print(f"Related papers identified: {summary['related_papers_found']}")
    print(f"Foundational papers (≤2015): {summary['foundational_papers']}")
    print(f"Recent developments (≥2020): {summary['recent_papers']}")
    
    print("\n🏆 Most Cited Papers:")
    for i, (pid, count) in enumerate(summary['top_cited_papers'], 1):
        title = network.papers.get(pid, {}).get('title', 'Unknown')
        print(f"  {i}. [{pid}] {title[:50]}... ({count} citations)")
    
    print(f"\n📁 Network exported to: {network_file}")
    print("="*70)
    
    return summary

# Execute review
results = conduct_comprehensive_review(
    seed_papers=["Seed_2020"],
    max_depth=2,
    output_dir="./my_review"
)

Expected Output Files:

literature_review/
├── foundational_papers.txt    # List of key historical papers
├── recent_papers.txt          # List of recent developments
└── review_summary.json        # Analysis statistics

Common Patterns

Pattern 1: Comprehensive Literature Review

Scenario: Starting a new research project and need complete background.

{
  "review_type": "comprehensive",
  "seed_papers": [
    "Most recent high-impact paper",
    "Recent comprehensive review"
  ],
  "depth": 2,
  "filtering": {
    "date_range": "2015-2025",
    "min_citations": 10,
    "journals": ["Nature", "Science", "Cell", "specialty journals"]
  }
}

Workflow:

Identify 2-3 recent high-quality seed papers
Run backward citation chasing (depth=2)
Run forward citation chasing (depth=2)
Identify top 20 most-cited papers
Filter by relevance and quality
Organize by themes/subtopics
Create reading schedule

Output Example:

Comprehensive Literature Review:
  Seed papers: 3
  Papers identified: 247
  After filtering: 85
  
Breakdown:
  - Foundational (pre-2015): 12 papers
  - Core developments (2015-2020): 38 papers
  - Recent advances (2020+): 35 papers
  
Themes identified:
  1. Methodological advances (23 papers)
  2. Clinical applications (31 papers)
  3. Basic mechanisms (31 papers)

Pattern 2: Foundational Paper Discovery

Scenario: New to a field and need to identify must-read classics.

{
  "task": "foundational_discovery",
  "approach": "citation_analysis",
  "seed": "Recent review paper",
  "metrics": [
    "citation_count",
    "citation_velocity",
    "author_h_index"
  ]
}

Workflow:

Start with recent comprehensive review
Extract all citations (backward chasing)
Rank by citation count
Check which papers are cited by multiple reviews
Read abstracts of top 20
Select 10-15 most relevant
Read in chronological order

Output Example:

Foundational Papers Identified:
  
Tier 1 (Essential - cited by everyone):
  1. Original discovery paper (2005) - 2,847 citations
  2. Methodology breakthrough (2008) - 1,923 citations
  3. First clinical application (2010) - 1,456 citations
  
Tier 2 (Important - cited by most):
  4-10. Major advances (2012-2018) - 500-1000 citations
  
Reading order: Chronological (2005 → 2018)

Pattern 3: Recent Development Tracking

Scenario: Staying current with latest advances in your field.

{
  "tracking_type": "recent_developments",
  "seed_papers": [
    "Seminal papers from 5 years ago"
  ],
  "direction": "forward",
  "depth": 1,
  "filter": {
    "year": "2023-2025",
    "alert_frequency": "monthly"
  }
}

Workflow:

Identify 5-10 key papers from ~5 years ago
Run forward citation chasing (depth=1)
Filter to last 2 years only
Sort by journal impact
Read titles and abstracts
Flag highly relevant papers
Set up alerts for key authors

Output Example:

Recent Developments (2023-2025):
  Papers citing key works: 127
  After year filter: 34
  
High-priority papers:
  - 3 papers in Nature/Science/Cell
  - 8 papers in top specialty journals
  - 12 methodological advances
  - 11 clinical studies
  
Key trends:
  - AI/ML integration increasing
  - Single-cell methods dominating
  - Clinical translation accelerating

Pattern 4: Interdisciplinary Bridge Finding

Scenario: Finding connections between your field and adjacent disciplines.

{
  "task": "interdisciplinary_mapping",
  "primary_field": "Your expertise",
  "target_fields": ["Field A", "Field B"],
  "method": "citation_bridge_analysis",
  "depth": 2
}

Workflow:

Identify top papers in your field
Run 2-hop citation network
Identify papers from other fields in network
Check cross-field citations
Map knowledge transfer pathways
Identify potential collaborators
Discover methodological opportunities

Output Example:

Interdisciplinary Connections Found:
  
Field A connections:
  - 12 papers cite your field's methods
  - 3 methodological papers highly relevant
  - Potential collaboration: Dr. X (3 shared citations)
  
Field B connections:
  - 8 papers apply your findings
  - 1 highly-cited review bridges fields
  - Conference overlap: ASM 2024
  
Bridge papers (cited by both fields):
  1. Methodology paper (2019) - 234 citations
  2. Application review (2021) - 156 citations

Quality Checklist

Pre-Analysis:

CRITICAL: Verify seed papers are high quality and relevant
Confirm seed paper IDs are correct (DOI, PubMed ID)
Check publication date of seed papers (recent vs. classic)
Verify access to citation database or API
Define scope clearly (topic boundaries)
Set citation depth appropriate for goals
Identify any papers to exclude (retractions, etc.)

During Analysis:

Track both directions (ancestors and descendants)
CRITICAL: Deduplicate papers found through multiple paths
Note citation context (meaningful vs. perfunctory)
Check for self-citation patterns
Verify chronological consistency
Flag review papers for special attention
Identify methodological papers separately

Post-Analysis:

CRITICAL: Filter results by relevance to topic
Rank papers by importance (citations, journal, recency)
Check for gaps (missing important papers)
Verify no key papers overlooked
Organize by themes or chronological development
Create prioritized reading list

Validation:

CRITICAL: Cross-check with domain expert or existing reviews
Verify foundational papers match field consensus
Check recent papers are actually novel
Ensure no important subtopics missed
Validate citation counts against Google Scholar/Web of Science
Review for potential bias (author, institution, country)
Document methodology for reproducibility

Common Pitfalls

Search Strategy Issues:

❌ Single seed paper → Limited coverage of field
- ✅ Use multiple diverse seed papers
❌ Only forward OR backward → Miss half the story
- ✅ Always search both directions
❌ Too shallow (depth=1) → Miss important connections
- ✅ Use depth=2 for comprehensive reviews
❌ Too deep (depth=3+) → Diluted, off-topic results
- ✅ Depth=3 only for historical analyses

Data Quality Issues:

❌ Ignoring citation context → Include irrelevant papers
- ✅ Read abstracts to assess relevance
❌ Counting all citations equally → Self-citations inflate counts
- ✅ Check for and exclude self-citations
❌ Not filtering by date → Outdated information
- ✅ Apply appropriate date filters
❌ Missing recent preprints → Not yet cited
- ✅ Supplement with bioRxiv/medRxiv searches

Interpretation Issues:

❌ Assuming more citations = better → Method papers over-cited
- ✅ Consider paper type and citation context
❌ Ignoring negative citations → Cited for being wrong
- ✅ Check citation sentiment when possible
❌ Confirmation bias → Only finding supporting papers
- ✅ Actively seek contradictory evidence
❌ Neglecting recent papers → Low citation count
- ✅ Use citation velocity, not just total count

Troubleshooting

Problem: No papers found

Symptoms: Empty results or "paper not found"
Causes:
- Incorrect paper ID format
- Paper not in database
- API connection issues
Solutions:
- Verify paper ID (DOI, PMID)
- Try alternative identifiers
- Check database coverage dates
- Use multiple search strategies

Problem: Too many papers to review

Symptoms: Thousands of papers identified
Causes:
- High-impact seed paper
- Broad field definition
- Too many seed papers
Solutions:
- Filter by date (last 5 years)
- Filter by citation count threshold
- Limit to specific journals
- Use depth=1 instead of 2

Problem: Off-topic papers included

Symptoms: Papers not relevant to your topic
Causes:
- Perfunctory citations
- Broad interdisciplinary citations
- Methodology papers cited everywhere
Solutions:
- Filter by keywords in titles
- Check citation context
- Remove papers with <N citations from seed
- Manual curation of results

Problem: Missing obvious papers

Symptoms: Important papers not in results
Causes:
- Wrong seed papers
- Incomplete database
- Recent papers not yet cited
Solutions:
- Add more diverse seed papers
- Supplement with keyword searches
- Check recent preprint servers
- Consult with domain experts

Problem: Circular citations

Symptoms: Papers citing each other (A→B→C→A)
Causes:
- Concurrent publication
- Cross-citation in reviews
- Methods validation
Solutions:
- Check publication dates
- Consider direction of influence
- Note concurrent development
- Manually resolve ordering

Problem: Self-citations dominate

Symptoms: Author's own papers most cited
Causes:
- Prolific author as seed
- Author reviews own work
- Series of related papers
Solutions:
- Exclude self-citations from counts
- Focus on external citations
- Flag self-citation patterns
- Consider author diversity

References

Available in references/ directory:

(No reference files currently available for this skill)

External Resources:

Web of Science: https://www.webofscience.com
Google Scholar: https://scholar.google.com
Scopus: https://www.scopus.com
PubMed: https://pubmed.ncbi.nlm.nih.gov
Semantic Scholar: https://www.semanticscholar.org
Citation Chasing Guide: https://guides.lib.uw.edu/research/citationchasing

Scripts

Located in scripts/ directory:

main.py - Citation network analysis and mapping engine

-- Node Attributes:

Size: Proportional to citation count
Color: Publication year (blue=old, red=recent)
Label: Paper ID or shortened title

Edge Attributes:

Direction: From citing to cited paper
Thickness: Can represent citation strength
Color: Can indicate citation type

Recommended Tools:

Cytoscape: Large networks, advanced analysis
Gephi: Visual exploration, force layouts
VOSviewer: Bibliometric maps, clustering

Last Updated: 2026-02-09
Skill ID: 187
Version: 2.0 (K-Dense Standard)

Parameters

Parameter	Type	Default	Description
`--doi`	str	Required	Paper DOI to query
`--title`	str	Required	Paper title to search
`--pmid`	str	Required	PubMed ID to query
`--network-file`	str	Required
`--demo`	str	Required	Run demo with example data
`--depth`	int	2
`--max-per-level`	int	20
`--delay`	float	0.5
`--output`	str	"citation_network.json"
`--format`	str	"json"

Citation Chasing & Mapping

Citation Chasing & Mapping

When to Use

Integration with Other Skills

Core Capabilities

1. Backward Citation Chasing (Ancestors)

2. Forward Citation Chasing (Descendants)

3. Foundational Paper Identification

4. Multi-Hop Network Analysis

5. Network Data Export

6. Field Evolution Mapping

Complete Workflow Example

Common Patterns

Pattern 1: Comprehensive Literature Review

Pattern 2: Foundational Paper Discovery

Pattern 3: Recent Development Tracking

Pattern 4: Interdisciplinary Bridge Finding

Quality Checklist

Common Pitfalls

Troubleshooting

References

Scripts

Parameters

Details