bibliometrix destekli bibliyografik eşleme

Bibliographic coupling links documents by the references they both cite (shared bibliography), making it ideal for mapping current research fronts because it reflects the authors' own framing of the literature at publication time. Co-citation links documents that are cited together by later papers, making it better suited to retrospective mapping of foundational knowledge structures. For a comprehensive picture, many studies run both analyses on the same corpus.

At least 100–200 documents with complete reference lists is a practical minimum. Below about 50 documents, the coupling network is too sparse for community detection algorithms to identify stable clusters. Very large corpora (5 000+ documents) should be pre-filtered by a minimum coupling strength to keep the network computationally tractable.

Yes. The `convert2df()` function accepts exports from Web of Science (plain text and BibTeX), Scopus (CSV), PubMed, and several other sources. When combining records from multiple databases, apply deduplication (e.g., using the `duplicatedMatching()` helper) before building the coupling network to avoid inflated similarity scores.

Bibliometrix defaults to Louvain, which is fast and performs well on large networks. Walktrap is an alternative that works better on smaller, denser graphs. The choice should be reported transparently; if the research question is sensitive to cluster boundaries, run multiple algorithms and note where results converge.

Yes. Biblioshiny is bibliometrix's Shiny-based graphical interface launched with `biblioshiny()` from R. It covers most coupling workflow steps through menus. However, Biblioshiny is less suitable for automation, custom normalisation, or integration with other analysis steps; for reproducible research pipelines, script-based usage is preferred.