Empowering Research Through Bibliometric Insights. biblio@vietscies.com Zalo 0916 086 008
Tools

VOSviewer Tutorial: From Installation to Publication-Ready Science Maps

A hands-on walkthrough of VOSviewer for keyword, co-authorship and co-citation maps built from Scopus or Web of Science data, plus the mistakes reviewers spot most often.

12 min read Updated 2026-11-21 Biblio VietSCI

VOSviewer is free software from the Centre for Science and Technology Studies (CWTS) at Leiden University, introduced in the methods paper by van Eck and Waltman (2010). It has become the default tool for science maps in bibliometric reviews, largely because it produces clean, readable networks with very little setup. This tutorial covers the whole workflow: installing the program, loading data, picking the right analysis, choosing thresholds, cleaning terms with a thesaurus file, reading the three visualization modes and exporting figures that meet journal requirements.

What VOSviewer does and when to use it

VOSviewer builds and visualizes bibliometric networks. Items (nodes) can be keywords, authors, organizations, countries, sources or cited documents. Links between items carry a strength, for example how many papers two keywords share. The VOS mapping technique places strongly related items close together, and a built-in clustering algorithm assigns each item to a colour-coded cluster.

It is the right tool when you want to:

  • Identify the main research themes of a field through keyword co-occurrence analysis.
  • Describe collaboration patterns among authors, institutions or countries.
  • Reveal the intellectual base of a field through co-citation of references.
  • Produce clear figures for the Results section of a review article.
Good to know VOSviewer excels at network visualization but offers little in the way of descriptive statistics. Many teams pair it with the bibliometrix package in R for productivity tables, citation indicators and thematic maps. See our Bibliometrix in R tutorial.

Installing VOSviewer

Download the program from the official site, vosviewer.com, under Download. At the time of writing (November 2026) the current release line is 1.6.x; check the website for the latest version before you start.

Step 1: Pick your platform

There are installers for Windows and macOS and a platform-independent version. The Windows and macOS installers normally bundle Java, so no separate Java installation is needed.

Step 2: Install to a simple path

Keep the program and your data in folders with short paths and no unusual characters to avoid file-reading errors.

Step 3: Plan for large datasets

Tens of thousands of records, or co-citation of cited references, can make the program slow. Close other applications and consider filtering your data before loading it.

Loading Scopus and Web of Science files

Click Create, choose "Create a map based on bibliographic data", then "Read data from bibliographic database files". VOSviewer reads the following formats directly:

DatabaseRecommended exportNotes
ScopusCSVInclude all fields, especially keywords and references if you plan a co-citation analysis.
Web of Science Core CollectionTab-delimited or Plain text, "Full Record and Cited References"Several files can be loaded at once.
Dimensions, Lens, PubMedFormats listed in the VOSviewer dialogNot every analysis type is available for every source.

VOSviewer cannot combine Scopus and Web of Science records in a single load. If you need both databases, convert, merge and deduplicate them first, usually in R. Our guide on exporting Scopus data explains how to prepare a clean file.

Choosing the type of analysis

After loading the data you choose a type of analysis and a unit of analysis. The table summarizes the combinations most often used in published reviews.

Type of analysisUnitsQuestion it answers
Co-authorshipAuthors, organizations, countriesWho works with whom; how international collaboration is structured.
Co-occurrenceAll keywords, author keywords, index keywordsWhich themes make up the field and how they relate.
CitationDocuments, sources, authors, organizations, countriesWhich units are cited most within the dataset.
Bibliographic couplingDocuments, sources, authors, organizations, countriesWhich recent papers share a common reference base.
Co-citationCited references, cited sources, cited authorsWhat the classic intellectual foundations of the field are.

You also choose a counting method. Full counting gives every link a weight of one; fractional counting splits the weight across the authors or keywords of each paper, which reduces the influence of papers with very long author lists. Our co-authorship and co-citation guide discusses this choice in detail.

Thresholds and the thesaurus file

Setting a minimum threshold

The threshold decides which items appear, for example keywords that occur at least five times. Set it too low and the map becomes a cloud of overlapping labels; set it too high and emerging topics disappear. A practical approach:

  • Start with the default and note how many items meet the threshold.
  • Adjust until the map holds roughly 50 to 150 keywords, or 30 to 100 authors or cited documents.
  • Record the final value and your reasoning so you can report it in the Methods section.

Building a thesaurus file

A thesaurus is a tab-delimited text file with two columns headed "label" and "replace by". Use it to merge variants ("covid-19", "covid 19" and "sars-cov-2 infection" into "covid-19"), to unify singular and plural forms, and to drop generic terms such as "study" or "analysis" by leaving the "replace by" column empty. Give it a clear name, such as thesaurus-terms.txt, and select it at the "Choose thesaurus file" step.

Tip Run the analysis once without a thesaurus, export the term list from the item selection table or the map file, and review it in a spreadsheet. Building the thesaurus from that list is faster and far more transparent than editing labels by hand.

Network, overlay and density views

Network visualization

Node size reflects weight (occurrences or citations) and colour reflects the cluster. Use it to describe the thematic structure.

Overlay visualization

Colour reflects a continuous score, usually average publication year or average citations. Use it to point to emerging topics.

Density visualization

A heat-map style view in item density or cluster density mode. Use it to show where research activity concentrates.

On the Analysis tab you can change the clustering resolution (default 1.00) and the minimum cluster size. A higher resolution yields more, smaller clusters. Keep these parameters consistent across maps if you intend to compare them.

Exporting high-resolution figures

Journals typically ask for at least 300 dpi or a vector format. In VOSviewer, use the Screenshot function, choose Save and increase the scaling factor before saving. Depending on the version, you can save raster images (PNG, JPG) and vector formats such as PDF, SVG or EPS; check the options in your current interface.

  • Adjust label size and the number of visible labels on the Visualization tab to prevent overlap.
  • Also save the map and network files (plain text) so the figure can be reproduced or edited later.
  • Write a complete caption: data source, time span, analysis type, threshold and counting method.

Common mistakes

Watch out Most problems come from the input data and the reporting, not from the software.
  1. Incomplete exports: references were not included in the Scopus export, so co-citation is impossible.
  2. No term cleaning: one concept is split across several nodes, which distorts the clusters.
  3. Unreported thresholds: reviewers cannot reproduce the map.
  4. Reading colours as rankings: cluster colours are labels, not an order of importance.
  5. Over-reading distances: positions are relative; you can say that nearby items are more strongly related, nothing more precise.
  6. Blurry figures: exporting at default resolution and enlarging the image in the manuscript.

Reading the map: a routine for interpretation

A finished map is a starting point for analysis, not a result in itself. A disciplined reading routine helps you write a Results section that reviewers can follow and check against the figure.

  1. Describe the overall structure first. How many clusters are there, how large is each, and is the network dense or made of loosely connected groups? VOSviewer reports items, clusters, links and total link strength; quote these numbers.
  2. Characterize each cluster. Sort the items of each cluster by occurrences or total link strength and read the top ten. Then read the abstracts of the most cited papers that contain them before you propose a label.
  3. Look at bridges. Items with strong links to several clusters often mark interdisciplinary topics. Hover over them to see their link strengths, or use the item list to compare them.
  4. Switch to the overlay view. Note which clusters sit at the recent end of the colour scale and which are older, and check these impressions against simple counts by year.
  5. Return to the data. Every claim you make about the map should be traceable to numbers in the map file or to specific papers, not just to how the picture looks.

Keeping a short log of these observations as you go also makes it much easier to answer reviewer questions months later, when the details of a particular run have faded.

When to bring in technical help

If you need to merge Scopus with Web of Science, clean thousands of keywords, or deliver journal-quality figures on a deadline, the Biblio VietSCI VOSviewer analysis service can take on the technical work: search strategy, data cleaning, thesaurus building, running the analyses, exporting high-resolution figures and handing over every source file so you can reproduce the results. Interpreting the clusters and drawing scientific conclusions remains the author's job. See our pricing (the Basic package, US$169, includes four VOSviewer maps) or send a request through the order page.

Frequently asked questions

Is VOSviewer free?

Yes. VOSviewer is free software developed by CWTS at Leiden University and can be used for research and teaching.

Can VOSviewer combine Scopus and Web of Science data?

Not directly. You need to convert, merge and deduplicate the two datasets first, typically with bibliometrix in R, before analysis.

What minimum number of occurrences should I use for keywords?

There is no universal value. Adjust the threshold until the map is readable, often 50 to 150 keywords, and report the value you chose in your Methods section.

What does a VOSviewer thesaurus file look like?

It is a tab-delimited text file with two columns headed label and replace by. Leave the replace by column empty to remove a term from the map.

How do I export a VOSviewer map at 300 dpi?

Use the Screenshot function and raise the scaling factor before saving, or save in a vector format if your version supports it. Check label sizes before exporting.

Ready to map the knowledge structure of your field?

Send us your research topic. We estimate the dataset size and recommend the right package free of charge within one business day.