Compute number of self citations with Semantic Scholar

self_cite.py

# License CC0 

import httpx

async def analyze_self_citations(doi):
    async with httpx.AsyncClient() as client:
        response = await client.get(
            f"https://api.semanticscholar.org/graph/v1/paper/DOI:{doi}",
            params={"fields": "title,authors,references.authors"}
        )
        response.raise_for_status()
        paper = response.json()

    if not paper['references']:
        print(paper)
        raise ValueError("Could not find references from Semantic Scholar")
    
    def check_overlap(n1, n2):
        # remove initials and check for name intersection
        s1 = {w for w in n1.lower().replace(".", "").split() if len(w) > 1}
        s2 = {w for w in n2.lower().replace(".", "").split() if len(w) > 1}
        return len(s1 | s2) == len(s1)
    
    authors = {a['name']: [0, 0] for a in paper['authors']}
    
    for ref in paper['references']:
        ref_authors = {a['name'] for a in ref['authors']}
        for author in authors:
            authors[author][1] += 1
            if any(check_overlap(author, ra) for ra in ref_authors):
                authors[author][0] += 1
    
    print(f"Self citations in \"{paper['title']}\"")
    print(f" N = {len(paper['references'])}")
    for author, (self_cites, total) in authors.items():
        print(f" {author}: {self_cites/total:.2%} self citation")

await analyze_self_citations("10.1038/s42256-024-00832-8")

# Self citations in "Augmenting large language models with chemistry tools"
#  N = 114
#  Andrés M Bran: 0.00% self citation
#  Sam Cox: 0.88% self citation
#  Oliver Schilter: 0.00% self citation
#  Carlo Baldassari: 0.00% self citation
#  Andrew D. White: 5.26% self citation
#  P. Schwaller: 6.14% self citation