Infrastructural Visibility and Language Mediation: A Mixed Methods Analysis of Metadata Asymmetry in Global LIS Publication Systems

This dataset accompanies the manuscript "Infrastructural Visibility and Language Mediation: A Mixed‑Methods Analysis of Metadata Asymmetry in Global LIS Publication Systems". It contains all data and code necessary to replicate the analyses reported in the paper. The repository includes: Raw bibliometric data from Global North (Scopus/Web of Science) and Global South (SciELO, SINTA, AJOL, Redalyc, ACI) sources. Cleaned and harmonized master datasets (CSV format). Jupyter notebooks (Python 3.12) for: Metadata harmonization and epistemic classification Journal concentration analysis (Gini, Lorenz) Topic modeling (LDA, coherence, Jensen‑Shannon divergence, cosine similarity) English‑only robustness check Logistic regression (English vs. non‑English) Supplementary figures and tables generation Output files: figures (PNG, 300 dpi), tables (CSV), and text reports. Supplementary materials (flow diagram, language distribution per index, homogeneity tests, etc.). All replication scripts use open‑source libraries (pandas, numpy, scikit‑learn, statsmodels, nltk, gensim, matplotlib). A requirements.txt file is included for dependency installation.