The 5-Step Pipeline
Every BibexPy analysis follows the same guided pipeline. Steps are gated — each unlocks when the previous one has produced its output — so a dataset can never skip preparation stages silently.
| # | Step | Output |
|---|---|---|
| 1 | Data & Merge | One consolidated, deduplicated dataset from raw WoS + Scopus exports |
| 2 | Records & Filtering | A reproducible, criteria-defined study corpus |
| 3 | Harmonization | Resolved author identities, consolidated institutions/countries, enriched metadata |
| 4 | Export | Analysis-ready files for VOSviewer, Biblioshiny, CiteSpace, reference managers |
| 5 | Report | Operation report + audit-grounded methodology narrative |
Analyses are isolated
Each merge run creates an isolated analysis inside the project. Re-merging after adding new raw files starts a fresh analysis while the old one remains intact in the project history — results stay reproducible even as the inputs evolve.
Everything is reversible
Before any operation that modifies records, BibexPy takes a snapshot. Every change is written to the append-only audit log with parameters and before/after states, and can be rolled back from the UI. See Report & Provenance.
Follow along
The next five chapters walk through each step with screenshots, using the built-in sample project.