VOSviewer is free software for constructing and visualizing bibliometric networks, developed by Nees Jan van Eck and Ludo Waltman at the Centre for Science and Technology Studies (CWTS), Leiden University. It is one of the two tools that account for most current bibliometric-mapping work in academic settings — alongside CiteSpace — and is typically the first tool a researcher, librarian, or research-information office reaches for when they need to turn a bulk export of publication records into a visual map of a field: clusters of related work, the researchers or journals at its center, and how its vocabulary or collaboration structure has shifted over time.
This guide covers VOSviewer specifically — what it is, its interface and workflow, its core use cases, and practical guidance for getting started. For the underlying methodology it implements (what co-citation and bibliographic coupling actually mean, and how a citation network is interpreted once built), see CASRAI’s Citation Network Analysis guide, which covers those concepts in depth; this page assumes that background and focuses on the tool itself.
What VOSviewer Is
VOSviewer constructs and visualizes bibliometric maps from several kinds of network built out of publication metadata:
- Co-citation networks — papers, authors, or journals frequently cited together by later work.
- Bibliographic coupling networks — papers, authors, organizations, or countries that cite the same earlier sources.
- Direct citation networks — straightforward citing/cited-by relationships.
- Co-authorship networks — collaboration structure between researchers, organizations, or countries.
- Keyword and term co-occurrence networks — terms that appear together in titles, abstracts, or author keywords across a corpus, extracted using VOSviewer’s built-in text-mining functionality rather than requiring the terms to already be tagged as keywords.
See CASRAI’s Citation Network Analysis guide for what co-citation and bibliographic coupling mean conceptually and why they were developed (Henry Small’s 1973 co-citation work and M. M. Kessler’s 1963 bibliographic-coupling work); this page treats those as known and focuses on how VOSviewer builds and displays the resulting network.
VOSviewer is developed and released by CWTS, the same Leiden University research center behind the Leiden Manifesto for research metrics and the CWTS Leiden Ranking. It is distributed free of charge for download and use, which is a significant part of why it is the most widely used bibliometric-mapping tool in published bibliometric-analysis papers: unlike a commercial analytics module bundled into a subscription database, it can be installed and run by any researcher regardless of their institution’s licensing, provided they can obtain the underlying bibliographic data to feed it.
Interface and Typical Workflow
VOSviewer runs as a downloadable desktop application (Windows, macOS, and Linux, via a Java-based build), and CWTS also offers VOSviewer Online, a web-based version for viewing and sharing interactive maps without a local install. A typical session follows a consistent sequence:
- Export or query data from a source database. VOSviewer accepts data drawn from Web of Science, Scopus, OpenAlex, Europe PMC, Semantic Scholar, Crossref, Lens.org, and plain RIS files, in addition to its own native VOSviewer JSON format. Newer versions also support querying OpenAlex directly from within the tool rather than requiring a separate export step first.
- Create a map from the imported data. VOSviewer’s “Create Map” wizard walks through choosing the network type (co-citation, bibliographic coupling, co-authorship, or co-occurrence), the unit of analysis (documents, authors, organizations, countries, or terms/keywords), and thresholds such as a minimum number of citations or occurrences a document or term needs to be included — important for keeping large, noisy corpora legible.
- Explore the resulting map. VOSviewer renders the network as a spatial map where item distance reflects relatedness (closer items are more strongly linked), color indicates cluster membership (or, in an overlay visualization, a variable such as average publication year), and item size reflects weight (typically citation count, link strength, or occurrence count).
- Switch between visualization types as needed — network visualization (nodes and links), overlay visualization (the same layout colored by a secondary variable like time), and density visualization (a heatmap-style view showing where item or cluster density is highest, useful for spotting a field’s dominant sub-topics at a glance without reading individual node labels).
- Refine and export. Cluster resolution, minimum link strength, and label size can all be adjusted interactively; the resulting map (or the underlying data table) can be exported as an image or file for use in a manuscript, report, or further analysis.
Core Use Cases
Mapping the structure of a research field
The most common published use: importing a large bulk export (often thousands of records) from a systematic search of a field, then running co-citation or bibliographic-coupling analysis to see how that field decomposes into sub-specialties, which works or authors sit at the center of each cluster, and how the field has shifted over time using an overlay visualization colored by publication year. This is the technique behind most standalone “bibliometric analysis” or “bibliometric review” papers.
Keyword and term co-occurrence analysis
VOSviewer’s text-mining module extracts noun phrases and terms directly from titles and abstracts (or uses author-supplied keywords), then maps which terms tend to appear together. This is used to surface a field’s dominant themes and how its vocabulary has changed, without depending on the accuracy or consistency of author-assigned keywords alone.
Co-authorship and collaboration mapping
Building a network of researchers, organizations, or countries based on shared authorship reveals collaboration structure — who works with whom, which institutions or countries are most central to a field’s collaboration network, and where isolated or bridging collaborations sit. This is commonly used in institutional research-strategy and research-office contexts, distinct from the literature-content mapping above.
Complementing systematic and narrative literature reviews
Increasingly used alongside (not instead of) a traditional literature review, to give a reviewer a structural, visual sense of a field before or during close reading — identifying major clusters and central works that a purely keyword-driven search might miss because of terminology drift across sub-communities. See CASRAI’s PRISMA and systematic review methodology guide for how a formal systematic review’s search-and-screen process itself works; VOSviewer is a complementary mapping tool, not a substitute for that documented process.
Getting Started: Practical Guidance
Data preparation matters more than the software
The quality of a VOSviewer map is entirely a function of the underlying export. A search strategy that is too broad produces a cluttered, uninterpretable map; one that is too narrow misses genuine sub-clusters. Before importing, it is worth deduplicating records (the same paper indexed differently across databases), deciding on a consistent unit of analysis, and setting a sensible minimum-occurrence or minimum-citation threshold so the map isn’t dominated by one-off noise terms or singleton documents.
Start with thresholds, not full detail
A first map built at a low minimum-occurrence threshold from a large corpus is usually unreadable. Raising the threshold (e.g., a term must occur in at least 10 documents, or a document must have at least 5 citations within the set) before evaluating cluster structure is standard practice, then lowering it selectively once the broad structure is understood.
Term co-occurrence needs manual cleanup
Because VOSviewer’s text-mining extraction is automated, it will surface near-duplicate terms (singular/plural variants, synonyms, abbreviations alongside their spelled-out form) as separate items. VOSviewer provides a thesaurus file mechanism to merge these manually before finalizing a map; skipping this step is one of the most common reasons a keyword map looks noisier or more fragmented than the underlying literature actually is.
Match the network type to the question
Co-citation is generally better suited to mapping a field’s established, cited-together intellectual structure; bibliographic coupling is better suited to finding papers addressing similar problems at roughly the same point in time, including very recent papers that haven’t yet accumulated citations themselves (since bibliographic coupling depends on a paper’s own reference list, not on how often it has been cited). Choosing the wrong one for the underlying question is a common source of a map that technically renders but doesn’t answer what was asked — see CASRAI’s Citation Network Analysis guide for the fuller distinction.
Cite the software and document parameters
Published bibliometric-analysis papers using VOSviewer typically cite the tool itself (van Eck & Waltman) and report the specific parameters used — source database, search string, date range, network type, and thresholds — so the map is reproducible. Because two analysts can build meaningfully different maps from the same underlying corpus depending on threshold and clustering-resolution choices, this documentation is what allows a reader to judge whether the map’s structure reflects the field or the analyst’s settings.
How VOSviewer Compares to Alternative Tools
VOSviewer is not the only bibliometric-mapping or citation-visualization tool, and different tools suit different tasks:
- CiteSpace — free, Java-based software from Chaomei Chen at Drexel University, focused specifically on detecting trends over time: research fronts, citation bursts (sudden spikes signaling an emerging topic), and structural turning points. It’s frequently used alongside VOSviewer in the same published paper, each covering a complementary analytical angle rather than one replacing the other.
- Gephi — a general-purpose, open-source network-visualization and analysis platform, not bibliometrics-specific. It offers more graph-layout and analysis flexibility than VOSviewer but requires the user to construct the underlying citation/co-occurrence network themselves (VOSviewer builds that network from raw bibliographic exports directly); researchers sometimes export a VOSviewer-built network into Gephi for more customized visual styling.
- Bibliometrix / biblioshiny (R) — an R package (with an optional web-app interface, biblioshiny) offering a much broader set of bibliometric indicators and statistical analyses beyond network mapping, at the cost of requiring more comfort with R or a scripting workflow than VOSviewer’s point-and-click interface.
- VantagePoint and BibExcel — older, more general-purpose text- and data-mining tools sometimes used for bibliometric data cleaning and analysis; BibExcel in particular was historically used to pre-process data before importing into a dedicated mapping tool, a role VOSviewer’s own text-mining and import features have largely absorbed for most current workflows.
- Web of Science and Scopus built-in analytics — both databases include native analysis dashboards (citation reports, author/institution comparisons, basic trend charts) that require no export or separate software, but are limited to each database’s own record set and don’t produce the same cluster-map or density visualizations VOSviewer does.
In practice, VOSviewer’s combination of being free, requiring no programming, and directly ingesting exports from major databases (including OpenAlex) is why it remains the default choice for most published bibliometric-mapping work, with CiteSpace as the most common complement rather than a straightforward substitute.
Frequently Asked Questions
Is VOSviewer free to use?
Yes. VOSviewer is developed by CWTS at Leiden University and distributed free of charge for download and use.
What file formats and databases does VOSviewer support?
VOSviewer imports data from Web of Science, Scopus, OpenAlex (including direct querying within the tool), Europe PMC, Semantic Scholar, Crossref, Lens.org, and plain RIS files, as well as its own native VOSviewer JSON format.
Do I need programming skills to use VOSviewer?
No. VOSviewer is a point-and-click desktop application (a web-based version, VOSviewer Online, is also available) built around wizard-style map creation; it does not require scripting, unlike R-based alternatives such as Bibliometrix.
What’s the difference between VOSviewer and CiteSpace?
Both are free bibliometric-mapping tools, but CiteSpace is built specifically around detecting trends and citation bursts over time, while VOSviewer is more general-purpose for co-citation, bibliographic-coupling, co-authorship, and keyword co-occurrence mapping. The two are commonly used together in the same published analysis rather than as substitutes for each other.
Can VOSviewer analyze keywords instead of citations?
Yes. VOSviewer’s text-mining module extracts terms directly from titles and abstracts (or uses supplied author keywords) and maps their co-occurrence, independent of any citation analysis, making it usable for thematic mapping even on a corpus where citation data isn’t the focus.
How is VOSviewer different from CASRAI’s Citation Network Analysis guide?
CASRAI’s Citation Network Analysis guide covers the underlying methodology — what co-citation and bibliographic coupling mean, how they were developed, and how a resulting network is interpreted — independent of any specific tool. This page covers VOSviewer itself: its interface, workflow, data-import options, and practical usage guidance.
