Reference Validator

2. Extract and Validate DOIs

For each reference in the file:

1. Extract the DOI using regex pattern: 10\.\d{4,}/[^\s\]\'">\}]+

2. Clean the DOI by removing trailing punctuation and whitespace

3. Validate via Crossref API:

   import requests
   doi = "10.1038/nature12345"  # Example
   url = f"https://api.crossref.org/works/{doi}"
   response = requests.get(url, timeout=10)
   is_valid = response.status_code == 200
   

4. Extract metadata if valid:

   • Title
   • Journal name
   • Publication year
   • Authors

3. Identify Problematic DOIs

Flag DOIs that are:

• Not found in Crossref (404 response)
• Annotation DOIs: contain "annotation" in the DOI
• Masthead DOIs: contain "masthead" in the DOI
• Preprints vs published: Check if preprint DOI when published version exists

4. Correct Invalid DOIs (Optional)

For invalid or suspicious DOIs, attempt correction:

1. Extract citation context (title, authors, journal, year)

2. Search Crossref by title and first author

3. Match with confidence scoring:

   • Score > 90: High confidence (safe to correct)
   • Score 70-90: Medium confidence (flag for review)
   • Score < 70: Low confidence (keep original)

4. Apply corrections only for high-confidence matches

5. Generate Reports

Create a comprehensive validation report with:

Summary Statistics:

• Total references processed
• Valid DOIs (count and percentage)
• Invalid DOIs (count and percentage)
• Problematic DOIs requiring manual review

Detailed Results: For each reference, include:

• Line number or reference ID
• Original DOI
• Validation status (valid/invalid/excluded)
• Corrected DOI (if applicable)
• Confidence score
• Title and journal (if found)

Recommendations:

• References needing immediate correction
• References requiring manual verification
• Suspicious patterns (e.g., multiple invalid DOIs from same journal)

Output Format

Corrected Reference File

Generate a new file with corrected DOIs while preserving the original format:

# References

1. **Smith, J. et al. (2023).** Correct paper title. *Journal Name*, 12(3): 123-145. DOI: 10.1234/correct-doi

2. **Jones, A. et al. (2022).** Another paper. *Different Journal*, 45(2): 67-89. DOI: 10.5678/also-correct

Validation Report

Create a detailed report (validation_report.txt) with:

====================================================================================================
DOI VALIDATION REPORT
====================================================================================================

SUMMARY:
Total references: 28
Valid DOIs: 24 (85.7%)
Invalid DOIs: 4 (14.3%)

====================================================================================================
INVALID REFERENCES REQUIRING ATTENTION:
====================================================================================================

Reference #4:
  Citation: Skarin, H. & Segerman, B. (2011). Horizontal gene transfer...
  DOI: 10.4161/mge.1.3.17714
  Status: NOT FOUND
  Action: Verify manually or remove from reference list

Reference #7:
  Citation: Smith, T. J. et al. (2021). The Distinctive Evolution...
  DOI: 10.3389/fmicb.2021.630537
  Status: ANNOTATION DOI
  Action: Replace with main paper DOI if available

====================================================================================================
CORRECTIONS APPLIED:
====================================================================================================

Reference #2:
  Original DOI: 10.1016/j.bbrc.2007.07.010
  Corrected DOI: 10.1016/j.bbrc.2007.06.166
  Confidence: High (score: 102.4)
  Action: Corrected automatically

[... more corrections ...]

Best Practices

Validation Strategy

1. Start with validation before attempting corrections
2. Use conservative thresholds for automatic corrections (confidence > 80)
3. Flag suspicious patterns like:
   • Multiple DOIs from same journal failing
   • DOIs with unusual formats
   • References without DOIs in otherwise complete list

Correction Guidelines

• Only apply high-confidence corrections (score > 90)
• Preserve original formatting of the reference file
• Always create a backup before making corrections
• Generate detailed reports for manual review

Quality Assurance

• Double-check corrections against the actual paper
• Verify journal names and publication years
• Check for duplicate entries with different DOIs
• Ensure consistency in DOI formatting

Common Issues

Issue: DOI Not Found in Crossref

Possible causes:

• Typo in DOI (wrong digit, missing character)
• DOI from predatory journal
• Reference to non-existent paper
• DOI format error (missing prefix/suffix)

Action:

1. Verify the DOI character by character
2. Search for the paper title in Crossref/PubMed
3. Check if the journal is legitimate
4. Consider removing the reference if verification fails

Issue: Annotation DOI Found

Example: 10.1371/annotation/b482f80f-c5b6-4b9c-8e9b-8b7139dc37f1

Action:

• These are usually corrections/addendums, not the main paper
• Search for the main paper DOI instead
• Verify which DOI is appropriate for citation

Issue: Multiple DOI Formats

Examples:

• 10.1016/j.bbrc.2007.07.010 vs 10.1016/j.bbrc.2007.06.166
• Same paper, different version (e.g., corrected article)

Action:

• Use the most recent/corrected version
• Verify which version matches the cited content
• Check journal's versioning policy

Advanced Features

Batch Processing

For large reference files (>100 references):

1. Process in batches of 50 to avoid rate limiting
2. Add delays between API calls (1-2 seconds)
3. Save intermediate results for recovery
4. Generate summary statistics after each batch

Confidence Scoring

Use Crossref's relevance score to assess correction confidence:

• Score > 100: Exact match (very high confidence)
• Score 80-100: Strong match (high confidence)
• Score 60-80: Probable match (medium confidence)
• Score < 60: Weak match (low confidence - do not auto-correct)

Exclusion Patterns

Automatically exclude certain DOI types from corrections:

• Annotation DOIs (contain "annotation")
• Masthead DOIs (contain "masthead")
• Component DOIs (contain "component")
• Dataset DOIs (unless citing datasets)

Troubleshooting

API Rate Limiting

If you get rate limited:

• Add delays between requests (1-2 seconds)
• Process in smaller batches
• Use exponential backoff for retries

Network Errors

For connection issues:

• Increase timeout values (10-30 seconds)
• Implement retry logic with backoff
• Check internet connectivity
• Use proxy if needed for institutional access

File Format Issues

If file parsing fails:

• Check file encoding (UTF-8 vs Latin-1)
• Verify DOI pattern matching
• Manually inspect problematic lines
• Consider pre-processing the file

Integration with Citation Styles

This skill works with various citation styles:

• Numbered: [1], 1., etc.
• Author-date: (Smith, 2023)
• Markdown: **Smith et al. (2023)**
• BibTeX: @article{...}

The key is extracting DOIs regardless of surrounding format.

Example Use Cases

Use Case 1: Pre-Submission Validation

User: "I'm submitting a paper to Nature and need to verify all my references are correct"

Action:

1. Validate all DOIs in the reference list
2. Flag any invalid or suspicious DOIs
3. Attempt corrections for high-confidence matches
4. Generate detailed report for final review

Use Case 2: Literature Review Quality Check

User: "I've compiled 200 references for my systematic review, want to make sure they're all real papers"

Action:

1. Batch process all references
2. Identify any fake or predatory journal papers
3. Check for duplicate entries
4. Generate quality report with statistics

Use Case 3: DOI Correction

User: "My supervisor said some of my DOIs are wrong, can you fix them?"

Action:

1. Extract all DOIs from the reference file
2. Validate each DOI via Crossref
3. Find correct DOIs for invalid entries
4. Apply corrections with confidence scoring
5. Generate before/after comparison

Important Notes

• Always preserve the original file - create corrected copies
• Use conservative correction thresholds to avoid false positives
• Manual verification is recommended for high-stakes submissions
• Keep detailed logs of all corrections made
• Consider the context - some fields have different DOI practices

Limitations

• Cannot verify references without DOIs (need additional verification methods)
• Dependent on Crossref API availability and accuracy
• May not detect all predatory journals
• Cannot verify paywalled content without access
• May have false positives for very recent publications

When to Escalate

Consider manual verification for:

• References to non-English journals
• Very recent publications (not yet in Crossref)
• References from obscure or predatory journals
• Government reports or technical documents
• Books and book chapters (may use different DOI systems)
• Conference proceedings with complex numbering