Reconstructing Journal Thematic Identity from Article-Level Topics: A Reproducible Framework and a Preregistered Temporal Validation

Journals are conventionally described through editorial subject classifications, yet what a journal actually publishes does not necessarily coincide with its assigned label. This work introduces a reproducible framework for reconstructing a journal's thematic identity directly from article-level evidence while maintaining a strict distinction between what is empirically observed and what is analytically inferred. From the distribution of OpenAlex research subfields across a journal's articles, we reconstruct an observed thematic representation and derive three preregistered comparative descriptors: alignment with the editorial classification, intrinsic coverage of the thematic nucleus, and relative competitive position. Together, these descriptors define an operational two-dimensional framework that separates a journal's intrinsic thematic identity from its competitive position within its competitive reference set. Using a transparently constructed cohort of journals, we characterize thematic concentration, temporal stability, and the discriminant structure of the proposed descriptors. The framework is then evaluated through a preregistered prospective replication that advances the observation window forward in time and through a documentary control that recomputes the same analysis using document-weighted rather than citation-weighted distributions. All analytical thresholds were fixed before analysis, and every reported result is linked to sealed preregistrations, reproducibility manifests, and cryptographically versioned artifacts. This is a preprint prepared for submission to Quantitative Science Studies and has not yet been peer reviewed. This Zenodo record constitutes the complete reproducible research package accompanying the submitted manuscript. It includes the canonical analysis code, sealed datasets, preregistration documents, reproducibility manifests, the complete Comparison D replication kit, cryptographic SHA-256 integrity manifests, and a code-to-documentation audit providing end-to-end traceability from every reported result to the exact implementation that produced it. The manuscript, documentation, and datasets are distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) License. Source code contained in the code/ directory is licensed separately under the MIT License.