S3 Bibliometric

Semantic Similarity Score (S3): a bibliometric method for mapping intellectual structures, packaged as a free Python library and an MCP server that LLM assistants (Claude, and any other MCP client) can call.

Give it a Scopus export. It finds the research communities of the field, their central papers, outliers, and the citation flows between communities. Your AI assistant can then label the communities and write up the intellectual structure.

How S3 works

1. Embed each paper's Title. Abstract with sentence-transformers/all-mpnet-base-v2
2. S3 matrix = cosine similarity between all papers, temporally masked (year_i ≤ year_j)
3. Scan thresholds 0.75–0.85: graph of S3 ≥ t → Louvain communities → silhouette
4. Keep the threshold with the highest SII = silhouette × clustered papers
5. Report communities, eigenvector centrality, outliers (papers with no S3 link), author keywords, TF-IDF terms and, when References + EID are exported, in-corpus citation flows with QI (within-community citations / size) and MI (share of field citations)

Louvain uses a fixed seed (default 42), so results are reproducible.

Use it from Claude (no install, no API key)

You need uv.

Claude Code

claude mcp add s3-bibliometric -- uvx s3-bibliometric

Claude Desktop: add this to claude_desktop_config.json (Settings → Developer → Edit Config):

{
  "mcpServers": {
    "s3-bibliometric": {
      "command": "uvx",
      "args": ["s3-bibliometric"]
    }
  }
}

Then ask, for example:

Run S3 on ~/Downloads/scopus_innovation.csv, label the communities, and write the intellectual-structure report.

You can also use the built-in s3_report prompt. The first run downloads the model (~420 MB) and PyTorch, so it takes a few minutes. Later runs are fast.

Runs are saved in ~/s3bib_runs (change it with the S3BIB_HOME environment variable).

Use it from Python

pip install s3-bibliometric

import s3bib

res = s3bib.run_s3("scopus.csv")       # or run_s3(df, s3bib.S3Config(start_year=2010))
res.scan                               # threshold scan
res.communities()                      # size, citations, years, QI/MI
res.community_papers(0, n=10)          # most central papers
res.top_keywords(0); res.top_terms(0)
res.outliers()
res.citation_matrix()
res.search("absorptive capacity")
res.labels = {0: "Open innovation", 1: "M&A"}
res.export("results/")                 # CSVs
res.save("runs/innovation")            # reload with s3bib.S3Result.load(...)

Input

Export from Scopus as CSV with at least Title and Abstract. Also used if present: Year, Cited by, Authors, Author Keywords, Source title, DOI, and References + EID (needed for the citation analysis).

Notes

• On a CPU, embedding 5,000 abstracts takes a few minutes. The silhouette step uses an N×N matrix (~100 MB at N = 5,000).
• A paper counts as citing another if the first 60 characters of the other paper's normalised title appear in its References. This step is O(N²).
• examples/make_sample.py creates a small synthetic dataset for testing.

Citation

If you use this package, please cite the S3 paper: Semantic Similarity Score: A New Bibliometric Method for Mapping Intellectual Structures in Business Research.

License

MIT