Getting Started

To get started with VOSviewer, you may want to take a look at some of the sources of information listed below.


Share this page

VOSviewer manual

The VOSviewer manual provides detailed and up-to-date information on the functionality of VOSviewer. The manual can also be accessed using the Manual button in VOSviewer.

Open VOSviewer manual


VOSviewer book chapter

VOSviewer is discussed in detail in a recent book chapter. The chapter provides a tutorial in which the use of VOSviewer is demonstrated in a step-by-step manner. The chapter is based on VOSviewer version 1.5.5. Some of the improvements and extensions introduced in the most recent versions of VOSviewer are therefore not covered in the chapter. In addition to VOSviewer, the chapter also discusses the CitNetExplorer tool for visualizing and analyzing citation networks of publications.

Open VOSviewer book chapter

VOSviewer paper

VOSviewer was originally introduced in a paper published in Scientometrics in 2010. In the meantime, VOSviewer has been improved and extended in many ways, so the paper is not fully up-to-date anymore. However, it still provides a useful introduction into the main ideas underlying VOSviewer.

Open VOSviewer paper

AIDA bookletAIDA booklet

The AIDA booklet introduces diverse case studies on research positioning and trend identification. These case studies are relevant for PhD candidates, researchers, group leaders, and policy makers. Many of the case studies make use of VOSviewer.

Open AIDA booklet


Mandatory fields
  • Syed Asif Raza April 20th, 2019 6:48 pm
    Dear Dr. Ludo,
    Thank you for comments.
    Regarding the co-citation matrix. Is there any video based (or document) tutorial for co-citation matrix creation.
    I noticed that while, I created co-citation matrix with Bibexcel, I got duplication references (paper) and many of these in the co-citation matrix. Is there any way to avoid this problem while using the VosViewer.
    Kind Regards,
    Syed A. Raza, PhD
    • Ludo Waltman April 21st, 2019 10:17 am
      In the case of a co-citation analysis, there may indeed be problems with multiple references that point to the same cited work but that differ in minor ways (e.g., in the way the source title is written). Using Web of Science, these problems tend to be relatively minor, but in the case of Scopus, these problems can be large. VOSviewer doesn't solve these problems, but it does offer the possibility of providing a thesaurus file as input. A thesaurus file can be used to solve the problem by manually specifying that certain references should be merged.
  • Mayo April 18th, 2019 12:53 pm
    Please how do i resolve this error.
    "Error while reading VOSviewer network file(line 2): Incorrect number of columns"
    Thank you.
    • Ludo Waltman April 20th, 2019 6:04 am
      Apparently the number of columns in line 2 in your network file is not correct, or at least it is not consistent with the number of columns in line 1 in your network file. Each line in a network file should have the same number of columns, either 2 or 3.
      • Mayo April 20th, 2019 7:16 am
        while i understand your explanation, this is a .csv downloaded directly from web of science. how do i troubleshoot it without tampering with the other data in the sheet.
        however, i have used the .txt and that worked perfectly fine for me.
        thank you.
        • Ludo Waltman April 20th, 2019 1:26 pm
          To my knowledge, CSV files cannot be downloaded from Web of Science, only text files. More importantly, the error message in your original message is about a network file, not about a Web of Science file. It therefore seems that you have chosen an incorrect option in the 'Create Map' wizard.
  • Howard Ramirez-Malule April 11th, 2019 8:26 pm
    Dear all,
    I have troubles uploading the CSV files exported from scopus. Do you have any suggestion for solving the problem?.
    IOException reading next record: (line 593) invalid char between encapsulated token and delimiter.
    Thank you very much in advance.
    Best regards,
  • Howard Ramirez-Malule April 11th, 2019 8:24 pm
    Dear all,
    I have troubles uploading the CSV files exported from scopus. Do you have any suggestion for solving the problem.
    IOException reading next record: (line 593) invalid char between encapsulated token and delimiter.
    Thank you very much in advance.
    Best regards,
    • Ludo Waltman April 13th, 2019 7:15 am
      Could you send your Scopus CSV file to We will then investigate your problem. Thanks!
  • Raysa Rocha April 11th, 2019 6:08 pm
    Good afternoon,
    I would like to know what the definition criteria are for choosing the 10 repetitions of the words in the full screen of a co-occurrence network map.
    I have 67 articles, the result with the default was totally different from when we decreased the amount of articles to 6.
    I would like an orientation on the length or lack of a co-occurrence amount.
    • Ludo Waltman April 13th, 2019 2:54 am
      Could you clarify your question? I don't fully understand your message.
  • Bárbara Cartagena da Silva Matos April 11th, 2019 3:09 pm
    Is it possible to import data from Excel spreedsheet to VosViewer? If yes, how?
    Thank you,
    • Ludo Waltman April 13th, 2019 2:48 am
      I am afraid this is not possible.
  • PJH April 4th, 2019 1:55 am
    Could you please provide a concise definition that could be used in an article to clarify the meaning of 'total link strength' as used in citation and co-citation analysis in VOSviewer. If the operational definition differs for the two types of citation analysis, please clarify that as well.
    • Ludo Waltman April 20th, 2019 3:49 pm
      Some definitions are provided in Chapter 2 in the VOSviewer manual. I'm happy to provide specific definitions for citation and co-citation analysis, but I then need to know your unit of analysis (e.g., documents, authors, or sources). Could you let me know your unit of analysis?
  • Alisa Mujkic March 31st, 2019 9:51 pm
    Dear dr. Ludo,
    please help us. :)
    We downloaded the collected papers available in Scopus for the journal we are interested in. However, there are additional 400 papers that are not available in Scopus, but in Science Direct, since the journal was not included in Scopus all the time.
    Is there any option to add those 400 papers in VOSviewer automatically or we have to go through this process manually? And please let us know which options we have to choose to do that in context of application itself?
    Best regards and thank you in advance,
    • Ludo Waltman March 31st, 2019 11:02 pm
      I'm afraid there is no easy way to add these 400 publications. I don't think it can be done automatically. I'm not sure how you want to do it manually. Manually adding publications to a CSV file exported from Scopus often causes problems.
      Perhaps there's an other data source (e.g., Web of Science or Dimensions) that indexes the 400 publications that you're interested in?
      • Alisa April 1st, 2019 9:04 am
        Dear dr. Ludo,
        thank you for your swift response and suggestion. However, other sources such as WoS or D do not have these papers - seems like the databases are “younger” then Scopus : S
        Finally, it turns out that the best possible solution for us is to change a topic.. :)
  • Sana March 30th, 2019 8:05 pm
    How to increase the memory? I cannot find JAR file as its not in download folder?
    • Ludo Waltman March 31st, 2019 9:46 am
      To obtain the JAR file, use the link 'Download VOSviewer 1.6.10 for other systems' on the 'Download' page on the VOSviewer website.
  • PJH March 29th, 2019 12:47 am
    A colleague asked me a question and I'd like to confirm if my response was accurate and complete. She asked"why do some authors shift between 2 clusters depending on the threshold I set?"
    I said, "Any author's co-cited publications will contain a distribution of documents, each of which has a different number of co-citations. Moreover, an author's publications may reflect different topical emphases. When you change the threshold, a different set of publications will be analyzed and the displayed clusters may shift. In these cases, however its likely that these shifts will be to an adjacent cluster. Since this isn't a case of 'right or wrong' select the cluster display that makes more sense in terms of the full set of relevant interpretation criteria "
    Is that explanation correct?
    • Ludo Waltman March 29th, 2019 6:45 am
      This explanation seems correct to me. In addition, when your colleague follows your advice, in her report I believe she should briefly explain the approach that she has taken, so she should explain that multiple thresholds have been considered and that the one that seems to give the most easily interpretable results has been chosen.
  • JK March 28th, 2019 2:01 pm
    Hi Ludo,
    I have a dillema in relation to author co-citation analysis in those cases when a paper is written by multiple authors.
    In VOSViewer manual I have found a following statement: „At the level of cited authors, author names extracted from the raw reference strings are used as the unit of analysis. When working with Web of Science or Crossref JSON files, only first authors are taken into account.“
    Does this mean that if I extract the data from SCOPUS in .csv format, names of all authors working on a paper will be taken into account in author co-citation analysis?
    Thank you.
    • Ludo Waltman March 28th, 2019 11:19 pm
      Yes, correct, in the case of a Scopus CSV file, all authors of a publication will be included in an author co-citation analysis.
  • Riffat March 25th, 2019 6:59 am
    i am working on VOS viewer by using web of science data while i run bibliometric analysis i use to run commend of open to vos map but it give me error that 100 line header is invalid how can i remove this?
    • Ludo Waltman March 25th, 2019 7:50 am
      Could you send your Web of Science file to I'll then take a look at your problem.
  • Syed A. Raza March 19th, 2019 10:41 am
    Dear Support staff,
    I like to ask about the co-citation analysis in VOSviewer. How we can export the citations (references) that are classified into clusters by software into exportable files (or we can copy them) such as txt, excel, file formats etc. for our analysis?
    Looking forward to hearing from you soon.
    kind Regards,
    Syd A. Raza, PhD
    • Ludo Waltman March 19th, 2019 5:07 pm
      You can use the 'Save' button on the 'File' tab for this purpose. Save your map in a VOSviewer map file. This is a tab-delimited plain text file that can for instance be opened in a spreadsheet program such as MS Excel. The information that you are interested in can be found in the map file.
      • Syed Asif raza March 23rd, 2019 7:14 am
        Dear Dr. Ludo,
        Many thanks for your assistance.
        Can you please also guide us, how to get the co-citation matrix from VOSviewer?
        Kind Regards,
        • Ludo Waltman March 23rd, 2019 10:56 pm
          In addition to a VOSviewer map file, you can also save a VOSviewer network file. The latter file includes the co-citation matrix.
  • Katie Hoemann March 15th, 2019 11:06 pm
    I am wondering if someone can help me understand how number of links and number of citations are determined for a document-level citation network. We're working with a list of 476 documents, one of which is cited by roughly 60-70 of the others (as determined by a string search through the input WoS file for the author, year, journal, and starting page). However, this document has only 6 links in the network. The same document has 491 citations, which exceeds the number of documents in the input data, but is far fewer than the number of citations given on WoS.
    • Ludo Waltman March 16th, 2019 7:28 am
      If you send your input file to, I will take a look at this issue.
  • Pedro March 12th, 2019 1:52 pm
    Dear Ludo,
    In a bibliometric network of countries, the diameter of circles are proportional to the number of publications. Their color represents which community that country has a major interaction with? How this interaction can be expressed/calculated? Based on the number of times they cite each other? Articles together? On other hand, the thickness of the lines is equal to the authors network (i.e., the thickness are proportional of the number of times they cite each other)?
    Thanks a lot for your answers!!!
    • Ludo Waltman March 12th, 2019 11:05 pm
      The answer to your question depends on the type of map your have created. Countries are clustered based on their relatedness. In the case of a co-authorship map, relatedness is determined by co-authored publications. In the case of a citation-based map, relatedness is determined by citation links.
      • Pedro March 13th, 2019 2:11 pm
        Dear Ludo,
        I have created a citation-based map. Then, the relatedness is determined by citation links (number of times they cite each other). Thanks a lot!!
  • Giuseppe Vito February 28th, 2019 9:42 am
    In my opinion, the potential of Vosviewer expresses itself at different scales, depending on the amount of examined data and the objective of the analysis. The chance of discussing a scientific map of a dimension -say 5000 references- is limited to a general level (think to an aerial photo taken from great height). A paper analyzing a limited amount of data (say > 25, as Ludo says, and < 200-300) is placed among the applications that allow a more detailed discussion. I hope I was helpful.
  • PJH February 27th, 2019 12:37 pm
    I just reviewed a paper for a journal which used VOSviewer and something in the paper raised a question for me. The database analyzed with VOSviewer was only 240 documents.
    Now I guess there is no best size, but I have tended to assume that larger databases get better results in terms of synthesizing trends. My databases have ranged from 700 to 25,000. I know some people look at databases in the millions.
    Is there a minimum size of a database needed to be able to obtain useful results? In this case I didn't reject the paper on the basis of size of the database, but I did wonder about it. Your thoughts?
    • Ludo Waltman February 27th, 2019 11:40 pm
      Too often I see VOSviewer visualizations based on rather limited numbers of documents. I am not always convinced of the usefulness of these visualizations. What constitutes a reasonable minimum number of documents depends on the type of visualization. For instance, a visualization of a citation network of documents may be highly insightful even if the data set includes just 25 documents. On the other hand, for a term co-occurrence map, I usually recommend to have at least 500 documents and preferably more.
      • PJH February 28th, 2019 12:04 am
        That makes sense was consistent with y own view. How about author co-citation maps?
        • Ludo Waltman March 1st, 2019 7:00 am
          I don't have a strong opinion on this, but I would tend to say that author co-citation maps don't require a very large number of documents. In the case of an author co-citation map, the authors are extracted not from the documents themselves, but from the reference lists of the documents. Hence, a single document will usually result in the identification of a substantial number of authors and a substantial number of co-citation links between these authors. Therefore, in order to have a meaningful author co-citation map, having just a limited number of documents may already be sufficient. Maps of authors based on direct citation or bibliographic coupling would typically require a larger number of documents.
          • PJH March 19th, 2019 12:26 pm
            Could you talk just a bit about what we can gain from a citation map vs. a co-citation map. I do understand the difference between these types of analyses.
            However, I am not clear about would we interpret an author 'citation map' as opposed to the author co-citation map.
  • Jenny Zhao February 26th, 2019 2:40 am
    Hello, I created a clustered network data with a similarity matrix and I created the map and network files, but I didn't specify the x and y coordinates of each node. The paper mentioned the method of automatically creating the network. Is it possible to display the data, knowing the similarity between two nodes without providing x and y coordinates?
    • Jenny Zhao February 26th, 2019 3:13 am
      Hey! I figured out one way to make it work. I just assigned random x and y values and then updated layout so it would create the x and y coordinates itself.
      • Ludo Waltman February 26th, 2019 7:17 am
        You actually don't need to provide x and y coordinates at all. Make sure to use the 'Create' button on the 'File' tab and to choose the option 'Create a map based on network data' in the first step of the 'Create Map' wizard. Do not use the 'Open' button on the 'File' tab if you do not have x and y coordinates.
        • Jenny Zhao February 27th, 2019 5:45 pm
          Hello again. I assigned the edge strength so that the larger the link strength the closer the nodes should be. However, I noticed that there is no relationship between strength and length. For example, out of three edges, one has the largest strength but the shortest length. The smallest strength has the middle length and the middle strength has the smallest length. Is there somehow to let vosviewer correlate strength and closeness?
          • Ludo Waltman February 27th, 2019 7:08 pm
            Distance and link strength should correlate, but the correlation usually is not perfect. This is because there are just two dimensions available. It is usually not possible to position items in a two-dimensional space in such a way that there is a perfect correlation between distance and link strength.
            Furthermore, the correlation in fact is not directly between distance and link strength but between distance and link strength normalized (by default) using the association strength method.
  • sivasundari February 25th, 2019 9:17 am

    Good Post! Thank you so much for sharing this pretty post,
    it was so good to read and useful to improve my knowledge as updated one, keep blogging.
  • Ana Hochmanova February 22nd, 2019 10:11 pm
    Good morning,
    I saved a map in a VOSviewer map file and for example got rows like this:
    label x y cluster weight weight
    league -0.647 0.5673 3 163 6.066.764
    hear 0.4082 0.629 2 191 32.014.138
    hair -0.1959 -0.5494 3 178 9.512.265
    I'm trying to have some measure of the quality of cluster 3, so I think that I can use the last column (weight).
    But I don't understand what does it mean the first one for example 6.066.764? or could I sum up all the weight for cluster 3= 163+178? would that be a measure of the quality?
    • Ludo Waltman February 24th, 2019 8:10 am
      I am not sure what you mean by the quality of clusters. Normally clusters do not have a quality. Could you clarify?
      • Ana Hochmanova February 24th, 2019 8:33 am
        Hello, I called it quality but I'm searching for a measure of how "good" a cluster is, how similar an object is to its own cluster (cohesion) compared to other clusters (separation). Some way to have an idea of intra-cluster similarity and inter-cluster differences.
        • Ludo Waltman February 24th, 2019 1:47 pm
          Thanks for clarifying. I am afraid that cluster cohesion cannot be determined based on the information provided in a VOSviewer map file. To determine cluster cohesion, you'll need to use a software tool or a library for network analysis. VOSviewer doesn't support this.
          • PJH February 24th, 2019 2:28 pm
            In the absence of measures that offer insight into "which map is better" I've been using a few guidelines:
            - use the tools available (thresholds, display #, cluster parameters) to try to get maps down to 4 or 5 clusters. I find this number easier to interpret. This tends to work for vo-citation maps.
            - using the same tools, I check different maps using tacit knowledge to judge fit and if the clusters make sense (a bit like factor analysis)
            - look for maps that show more coherent as opposed to overly interminglling clusters (thought I don't discount all overlapping since that can also 'make sense'.
            Ludo should feel free to say I'm all wrong but thats where Ive ended up after working on a dozen papers with VOSviewer.
            Final note: I've had reviewers complain about the lack of statistical parameters to help define an optimal map. But it hasn't been insurmountable.
            • Ludo Waltman February 24th, 2019 3:42 pm
              I like your advice, especially the point on using tacit knowledge. This is also what I do myself. It makes sense because the maps should be used primarily as an exploratory (rather than a confirmatory) tool. I am familiar with the complaint about lack of statistical measures, for instance measures of goodness of fit or something similar. I think the idea of trying to quantify everything is problematic. What matters is whether a map is useful for a particular purpose, and usually there is no clear way to quantify this. This is something we should judge qualitatively.
              • PJH February 24th, 2019 11:04 pm
                Well you made my morning Ludo, to know that I have been on the right track! hahaha
                Ana, although Ludo was unsure at first what you meant, I got it immediately since I had also struggled with the assumption that there would be some parameters. However, a dozen papers later, I have come to realize that this tool gives you a quantitatively derives empirical measure of construct patterns that we previously could only guess at. And only scholars with deep experience could even guess. However, the "interpretation" of the patterns and even determining which set of patterns (e.g., which map) is 'best' must be made using your prior knowledge of the field. This has 2 implications.
                First, if you are a doctoral student or someone new to the field or line of inquiry that you are analyzing, it will be more difficult to make these judgments since you are 'low' on tacit knowledge. The good news is that this is also a great way for those scholars to get 'up to speed' in a field because in order to make those judgments you have to examine some selection of the papers or authors comprising a cluster.
                Second, if you are already very familiar with the field, the 'sense-making process' will be quite fast when looking at the maps. The only trick is to be sure not to pick the first map that fits to your preconceptions, but rather to examine a number of different maps using different parameters. Actually, in this case the clusters tend to change at the margins rather than in major ways (e.g., 4 vs 5 or 6 clusters).
                In conclusion I now find the use of 'qualitative methods of interpretation' of a quantitatively derived set of patterns quite liberating! Good luck and enjoy.
                • Ana Hochmanova February 25th, 2019 7:53 am
                  Thanks a lot for all your comments, I'm in the first group :)
                  I'm trying to reduce the number of clusters, in fact I already have 4 or 5 that can actually be interpreted.. but also several "small" clusters with 2 to 8 elements that I don't know what to do with them..
                  You might think I'm a bit stubborn :) but if I still needed to give some kind of numerical measure to clusters.. could I use the last two fields in the map file? the Link and total link strength. "For a given item, the Links and Total link strength attributes indicate, respectively, the number of links of an item with other items and the total strength of the links of an item with other items".
                  If those two are high values with small standard deviation wouldn't that be an indication of a high cohesion?
                  • Ludo Waltman February 25th, 2019 8:57 am
                    This would not be a good idea. Based on the information in the map file, you cannot distinguish between intra-cluster and inter-cluster links. Therefore the information in the map file doesn't tell you anything about the cohesion of clusters.
                    • Ana Hochmanova February 25th, 2019 9:04 am
                      That is because the Total Link strength indicates the total strength with any item, not with the items in the same cluster?
                      • Ludo Waltman February 25th, 2019 9:06 am
                        Yes, correct.
                    • PJH February 25th, 2019 9:04 am
                      The way to manage this Ana was previously suggested by Ludo. When you have co-citation or co-word map open, click the tab on the upper left menu "analysis'. Then you have two options under "Clustering":
                      1. For your issue you could just change the value in the "minimum cluster size" box to whatever works best (e.g., 3, 6, 8).
                      2. Another alternative is to change the "resolution" to .9 or .8 and see what happens.
                      Hope this helps. I am no expert, just a seeker like yourself. However, trust me, persevering will be worth the effort. :)
  • Ilenia Bregoli February 20th, 2019 7:33 pm
    Dear Ludo and team,
    I finished my search on Scopus of the articles that I want to analyse. I saved them in a list in Scopus and now I downloaded the file as a .csv. However, I am not able to see any analysis because once I upload my file there is the message: "Error while reading Scoups file "file name" (line 11): incorrect number of columns". I do not understand why this message appears since I have not opened the file, I just saved it and uploaded in Vosviewer. Could you please help me in sorting this issue out?
    I thank you in advance for your help.
    Best regards,
    • Ilenia Bregoli February 20th, 2019 11:55 pm
      it's still me. I just wanted to tell you that I sorted the issue. Apparently there was a problem with the funding details of some papers. Once I have not downloaded those details, now the software works fine :-)
      Best regards,
      • Ludo Waltman February 20th, 2019 11:58 pm
        Yes, you are right, there is indeed a problem with funding data in CSV files exported from Scopus. For more information, please see the box 'Problems with Scopus data' on
    • Silva February 21st, 2019 3:08 pm
      I'm having the same error when importing in Scopus file csv.
  • Atefeh Zolfagharnasab February 19th, 2019 10:26 am
    Dear Ludo
    Creating a map based on text data is difficult because the disparity of the various forms of the words is high.
    Is there a way to convert and merge words into their root before creating a map based on text data?
    • Ludo Waltman February 20th, 2019 11:55 pm
      I am not sure whether I understand your question. VOSviewer takes into account only nouns and adjectives. Nouns are converted to their root by transforming plural nouns into singular ones. I should note that this works only if your text is in English, which is an important requirement when providing text data as input to VOSviewer.
      • Atefeh Zolfagharnasab February 21st, 2019 10:00 am
        I mean, for example, all the words that are related to spiritual health, such as spiritual care or spiritual healing, are in the collection of spirituality.
        • Ludo Waltman February 21st, 2019 10:26 am
          Thanks for the clarification! To address this issue, you need to make use of a thesaurus file. Please see the VOSviewer manual for more information.
          • Atefeh Zolfagharnasab February 22nd, 2019 6:56 am
            Thanks a lot.
  • Tomaszewski February 17th, 2019 9:14 pm
    any way of deciding which method for normalization to use (eg. association strength or fractionization or no normalization?
    • Ludo Waltman February 20th, 2019 11:52 pm
      For ordinary users of VOSviewer, I recommend to always use the association strength. The other normalization methods are intended to be used only by advanced users and only in special cases.
      • Robert Tomaszewski February 22nd, 2019 7:03 am
        Thank you
  • Xingdao He February 17th, 2019 12:07 pm
    Dear friends,
    a have another question.
    the first example in this web is a map of scientific fields, but i can't find the button to creat a VOSviewer map about scientific fields in the software.
    i'd be appreciate if you coud tell me how to creat a map about scientific fields.
    • Ludo Waltman February 20th, 2019 11:38 am
      I am afraid there is no easy way to create this map. The map was created by first preprocessing a large amount of Web of Science data and then providing the preprocessed data as input to VOSviewer (in a VOSviewer map file and a VOSviewer network file). This can be done only if you have large-scale access to raw Web of Science data.
      • Xingdao He February 20th, 2019 2:21 pm
        thanks for your reply.
        i have already had a large amout of Web of Science data, but i don't know how to creat it. can you gave me some detailed courses or study materials about how to creat a scientific fields map?
        tnanks a lot!
        • Ludo Waltman February 20th, 2019 11:50 pm
          I am afraid I cannot easily help you with this. At my center, we have a relational database environment in which we work with Web of Science data. In this environment, we use SQL queries to create all kinds of bibliometric networks, such as a citation network of scientific fields. These networks can then be provided as input to VOSviewer.
  • Xingdao He February 17th, 2019 12:01 pm
    Dear friends,
    in the author citation map, the two author are very close shows that they may be partner in one lab or be in a small field, why two authors have a very close distance but are assigned to different clusters?
    • Ludo Waltman February 20th, 2019 11:43 am
      There can be different reasons for this. For instance, suppose we have two researchers who belong to different communities but who nevertheless work quite regularly on the same topics. The fact that the researchers are assigned to different clusters could reflect that they belong to different communities, while the fact that they are located close to each other could reflect that they work on the same topics. Furthermore, keep in mind that the clusters depend on the choice of the resolution parameter in VOSviewer. By lowering the resolution parameter, the number of clusters can be decreased, and it is likely that the two researchers will then be assigned to the same cluster.
      • Xingdao He February 20th, 2019 2:29 pm
        thanks a lot! i've got that.
  • Fabiola Andrade February 13th, 2019 2:39 am
    Dear Ludo and Team,
    I tried to build an. xlsx file with the same basic items as the files generated by the WoS platform (publication type, authors, references, keywords, titles, etc.), so that you can later apply the. txt file to VOS. However, when I insert the latter, the software displays the following message (which prevents me from continuing the mapping): "Error while reading the Web of Science file (file name) (line 25): Incorrect number of columns". Is there anything that you or I can do to generate the maps?
    • Ludo Waltman February 13th, 2019 7:54 am
      This error message means that your file doesn't properly follow the WoS file format. It is not so easy to create your own files in the WoS file format. This often causes problems for VOSviewer users. If you send your file to, I could take a look at it to see if I can give any further advice.
  • Pedro February 12th, 2019 4:25 pm
    Dear friends,
    In software, if I set the type of analysis as "Citation", and the unit of analysis as "Authors", and set up the "Overlay Visualization", I will have an authors' network. I can set, as example, the scale of circles as proportional to the number of documents as the weight and the Avg. pub. of year as the score. Thus, about the total link strength: it will indicate the frequency that the authors were cited together? Or it will be proportional to how they cite each other?
    • Ludo Waltman February 12th, 2019 7:01 pm
      For two authors A and B, the link strength equals the number of citatons given by A to B plus the number of citations given by B to A. The total link strength of an author A equals the sum of the link strengths of A with all other authors.
      • Pedro February 12th, 2019 8:01 pm
        Thanks Ludo.
  • Pablo Rodriguez February 12th, 2019 10:14 am
    Good morning, I'm trying to create a map following the steps:
    Create a map based on network data - Select files - Chose threshold -
    i have 250 items and in the Select items I choose all of them but when I create the map only about 10 items can be seen because they are all one above the other, I mean that all items are shown so close to each other that they supperimpose so I can see only about 10 of the 250.. and no clusters can be seen, only a lot of small circles of different colors all toguether one above the other only occuping about 5% of all the screen..
    I mean the screen all white and all the items colapsed toguether ocupping a very small part of the screen.. I've been trying to expand it somehow but can't find how to do it...
    • Ludo Waltman February 12th, 2019 3:58 pm
      The best way to address this issue is by lowering the value of the repulsion parameter on the 'Analysis' tab. For most analyses, the default value of this parameter is 1. You could for instance lower the value to 0, -1, or -2.
      • Pablo February 15th, 2019 10:59 am
        Thanks a lot Ludo it worked fine!!
  • Pu Chen February 11th, 2019 7:57 am
    Dear inventor,
    Do you need to cite any specific article for freely using VOSviewer in a publication purpose?
    • Ludo Waltman February 12th, 2019 3:55 pm
      Thanks for asking. You don't need to cite anything. However, many researchers that use VOSviewer choose to cite the VOSviewer paper listed at the top of this page, or alternatively they cite the book chapter about VOSviewer that is also listed above.
  • Mooyeon KIm February 5th, 2019 10:11 am
  • Atefeh Zolfagharnasab February 2nd, 2019 7:33 am
    hello. I'm working on text mining project. data was exported from Mendeley on RIS format. I use "create a map based on text data". Before creating map step, when the list of words is displayed for selection, I need to have these words in a MS Word file to present them to expert to decide how to choose words. How can I do this?
    • Ludo Waltman February 3rd, 2019 10:26 pm
      The list of terms can be exported by right-clicking on the list. A context menu will appear that offers an option to export the list of terms. Using this option, the terms will be exported in a text file that can be opened in for instance MS Word or MS Excel.
      • Atefeh Zolfagharnasab February 4th, 2019 7:19 am
        Hello. Thank you all the same, I'll try with it.
  • Andréia Anschau February 1st, 2019 4:24 pm
    Hi, I have doubts about the construction of the journal network map in which the scale in the “overlay visualization” in relation to the impact factor of the journal (as in Figure 3, page 11 of Manual for VOSviewer version 1.6.10). Until now I only obtain this map with the scale in relation to the year of publication. From what I understand, two documents are needed, for example the "journal_map.txt." and "journal_network_sparse.txt." templates. I get my data similar to the "journal_network_sparse.txt.", but I am doubtful about the data to be used for my "journal_map.txt." Is it necessary to use all the information contained in this template? I tried to use a "journal_map.txt." containing only the ID and score , however an error message appears in line 1 "error while reading VOSviewer network file (line 1): invalid ID”. The IDs I entered in my "journal_map.txt." were extracted from the last step of the 'Create Map' wizard, saving the links between terms and documents. Can you give me an orientation about this? Maybe send me a "journal map" model with only the information I need for this type of map or a spreadsheet data so I can visualize better
    • Ludo Waltman February 3rd, 2019 11:33 pm
      Maps such as the one presented in the VOSviewer manual cannot easily be created. There is no easy way to create these maps based on data downloaded through the Web of Science online interface. The map presented in the manual was created based on data extracted from the in-house version of the Web of Science database available at CWTS. The impact factor data was obtained from this database. I am afraid this data cannot easily be obtained through the Web of Science online interface.
  • Anton January 30th, 2019 9:50 am
    Hi Ludo,
    Really appreciate your wonderful work and the commitment to answering various queries.
    I have an unresolved issue in making the maps available online. I have followed the exact steps in the manual, however, when downloading and opening the files, i get an error: "Error ...: File cannot be found".
    I have made the map available to public via Google Drive as well as direct download. Neither require passwords. When clicking the link the software says "Error ...: File cannot be found". However, the link (for the map) itself seems to work and download fine: (direct download link) or to open with the browser. Alternatively, if i try to input the following address (without the direct download file): it says: "Unable to create map: Incorrect use of quotes".
    I'm struggling with this issue for a while now, is there anything you could suggest, please?
    Kind regards,
    • Ludo Waltman February 3rd, 2019 11:38 pm
      This problem can be solved by replacing the ampersands after ‘download’ by the code ‘%26’.
      • Anton February 4th, 2019 7:52 am
        Ludo, you are a legend. Thank you, so much!
  • Andy Yeung January 24th, 2019 10:39 am
    Hi everyone, is it possible to select some bubbles not to be visualized, by altering the files of map.txt and network.txt. My computer spent a long time to digest the large dataset and generate the term map / keyword map, so I do not want to re-run the processes again just to remove bubbles that are not relevant for my visualization purpose.
    • Ludo Waltman January 24th, 2019 6:20 pm
      Yes, this is possible. You simply need to remove the relevant lines from the map file. The lines in the network file that point to the items to be excluded from the visualization need to be removed as well.
  • MURENDENI LIPHADZI January 16th, 2019 9:36 am
    I can seem to run vosviewer on my mac..i get a message saying unable to run java run time environment everytime i open the application. please help
    • Rafael L February 15th, 2019 11:01 pm
      I'm getting the same error message. Have you found out the solution?
  • Raquel Reis January 14th, 2019 1:29 pm
    I have a network database in pajek format, but I'm having trouble importing it into VOSviewer. The problem is that when I import the file containing the clusters, the error appears that there is no identification of the vertices. I would like to know how the base format should be (what it should contain in rows and columns) so that the program can read my base in pajek or excel format. I did not find this information in the manual.
    • Ludo Waltman January 14th, 2019 4:22 pm
      Could you please send your files to, so that we can take a further look at your problem?
  • Philip December 24th, 2018 11:01 pm
    When I run citation analysis, the results seem to be unstable.
    Please advise:: Should the results with respect to citation totals differ depending on the thresholds?
    I would think that total citations should be a stable total if one is using the same database.
    • Ludo Waltman December 24th, 2018 11:45 pm
      Could you please provide a bit more details: Which type of citation analysis did you carry out (i.e., citations, bibliographic coupling, co-citation)? What is your unit of analysis (e.g., documents, sources, authors)? What do you mean by 'citation totals'?
      • Philip December 24th, 2018 11:51 pm
        Citation analysis of authors and citation of documents Total citations.
        When I rerun the anal.ysis at different times, I get slightly different rank order and some different citation results. I thought I was using the same settings, but I didnt thing the number of citations would change with the settings.
        • Ludo Waltman December 24th, 2018 11:54 pm
          What do you mean by 'citation of documents Total citations'?
          • Philip January 30th, 2019 7:45 am
            Sorry for the delay in responding. I am running "citations of documents" on a database of 1,842 documents. I am using a thesaurus file.
            Case 1: When I set the minimum citation threshold at 1 or 2 citations, I get a list in which the top cited docs are 250/189/181.
            Case 2: When I set the minimum citation threshold at 3, 4, 5, or 6, I get a list in which the top cited docs are 522, 250/189/181 (doc with 522 is a new doc which did not appear in Case 1).
            Case 3: When I set the minimum citation threshold at 7 citations and higher, I get a list in which the top cited docs are 522, 250/250/189/181 (2nd doc with 250 is a new doc which did not appear in Case 1 or 2).
            I have run into this kind of problem before. I used to assume it was my error and just kept changing my tables. However, it is something that seems to happen frequently.
            What am I missing here?
            • Ludo Waltman January 30th, 2019 10:55 am
              There are two types of citation statistics in VOSviewer. On the one hand, there is the total number of citations a document has received in the entire Web of Science database (or whatever database is used). On the other hand, there is the number of citation links a document has within the current data set provided as input to VOSviewer. I think the distinction between these two concepts is causing confusion. When you set a threshold based on the number of citations of a document, this threshold applies to the total number of citations received in Web of Science. When you look at the number of citation links of a document, this is the number of links within the current data set.
  • Marina December 24th, 2018 5:08 pm
    Hello, I tried to open my files but an error returned
    "Error while reading web of science file..... is not a valid column header" What I did wrong?
  • Andy Yeung December 23rd, 2018 3:53 pm
    Thanks for developing the software and updating it from time to time. At times I opened the map.txt for term map, and did notice that the Occurrence count was smaller than the number returned by WoS on manual search. Same situation applied to map.txt for citation map of Organization or Country. Is it because VOSviewer would discard some records with missing info, e.g. those without abstract?
    On the other hand, would it be possible to add a function that can list the documents responsible for the specific terms displayed in the term map? Or at least to give 95% confidence intervals for the averaged citation count data. Because the colored bubble maps are very eye catching, but reviewers and readers may want to know if the colors are stable (with small 95% C.I.) or influenced by outliers. Thanks for your attention.
    • Ludo Waltman December 23rd, 2018 4:06 pm
      In the final step of the 'Create Map' wizard, a table is presented listing the items (e.g., terms, organizations, or countries) that will be included in your map. It is possible to right-click on this table. A context menu will then appear that offers an option to export the links between the items and the documents in which the items occur. By exporting these links, you know for each item the underlying documents, making it is possible to answer questions such as the ones you have.
      By the way, VOSviewer does not discard any documents. In the case of organizations and countries, I am not sure why statistics reported by WoS differ from those reported by VOSviewer. In the case of terms, this might be due to the way in which term occurrences are counted by VOSviewer. For instance, if a term 'neural network' occurs in a document, VOSviewer will count this as an occurrence of 'neural network', not as an occurrence of 'network'. This might explain the difference between statistics provided by WoS and VOSviewer.
  • Pablo Rodriguez December 8th, 2018 11:40 am
    Thanks a lot for this incredible tool! I just wanted to ask if is possible to export the items after clusters are generated, in the Items tab we can see for each cluster the items that conform them, is possible to export such list of items?
    Also similarly, when we create the map for example "create a map based on network data" then I select the map file and the network file, then I have to chose the total link strength of an item and at last there is a list of all items and their respective strengths, is possible to export or select or copy that list of items with their strenghts?
    Because I think it is important to show the strenght of each item as a measure of the "quality" of the item, if I understand it right the larger the strength the more important is the selected item.
    Thanks again for your great work
    Best regards
    • Ludo Waltman December 8th, 2018 12:28 pm
      When you save a map in a VOSviewer map file, the information that you are looking for can be found in this file. A map file is a plain text file with tab-delimited columns. It can best be opened using a spreadsheet program such as MS Excel.
  • giuseppe vito November 19th, 2018 10:25 am
    Dear Ludo, dear Philip H., I had encountered the problem having already worked uncheking FOUNDING DETAILS and "Tradenames & manufacturers" and "Accession numbers & chemicals" in the section OTHER INFORMATION. Uncheking in BIBLIOGRAPHIC INFORMATION everything except "Affiliations" and "EID", "Access Type" and "Author (s) s ID" in the section CITATION INFORMATION, everything works for the purpose of my work. The consistency remains in the fact that with the same restrictions on the export of metadata, some subsets of a research on Scopus that is treated by Vosviewer without problems, are not read by the program. I hope to have contributed to the issue, also suggesting something to the Scopus team. Best regards
  • giuseppe vito November 18th, 2018 6:05 pm
    Dear Ludo, here I am again:
    a csv file from Scopus obtained with the query (SRCTITLE (sustainability) AND AFFILCOUNTRY (italy)) AND (LIMIT-TO (DOCTYPE, "ar")) returns 900 refs. [of which 42 contain the keyword "climate change"] and Vosviewer treats it without problems.
    By limiting the search in these 900 docs to the exact keyword "climate change" with the query
    (SRCTITLE (sustainability) AND AFFILCOUNTRY (italy)) AND (LIMIT-TO (DOCTYPE, "ar")) AND (LIMIT-TO (EXACT KEYWORD, "Climate Change")) Scopus returns the csv file containing 42 refs but Vosviewer fails to read it by returning the error "... incorrect number of columns". The fact amazes me because this second file should be (is) a subset of the first. Can you kindly help me?
    • Ludo Waltman November 18th, 2018 6:35 pm
      This is indeed very strange, but I am afraid I cannot help with this. There currently is a problem with the Scopus CSV export. The nature of the problem is not entirely clear to me. The problem needs to be fixed by Elsevier's Scopus team. Hopefully they will do so as soon as possible. There is not much I can do at the moment.
      • giuseppe vito November 18th, 2018 6:48 pm
        Dear Ludo, thank you all the same, I'll try with Web of Science. Good Sunday evening
        • Philip H. November 18th, 2018 11:24 pm
          Please that the workaround offered previously by another contributor here DOES WORK. In the selections under CSV Export UNCHECK the box next to FUNDING DETAILS. Then try to export and upload.
          It has worked for me even after I ran into the same exact problem.
      • Maryam Mazaheri November 23rd, 2018 11:24 pm
        Ludo Waltman
        As i had the same problem with scopus, i try publish or perish software and the Csv. result File that the software get is shown in vosviewer.
  • Nikhil November 16th, 2018 12:50 pm
    Hello everyone, Can someone please clarify?
    When I create a co-citation(cited references) map, a particular node has following information:
    Item: graham Jr, 2005, i account econ
    cluster: 3
    links: 215
    total link strength: 739
    citations: 47
    If I understood correct this node has 215 links with different nodes in the map and the sum of strengths of all links is 739. But what is do this 215 links actually show? is it no of co-citations? if not how can I find no of co-citations?
    Thank you in Advance
    • Ludo Waltman November 16th, 2018 10:58 pm
      When you move the mouse over a co-citation link, you can see the strength of the link in the status bar. The strength of the link equals the number of co-citations of two cited references.
      • Philip H. November 17th, 2018 2:12 am
        Could you please clarify this point further concerning 'strength of links'.
        - I had thought that the lines 'linking' two nodes on a citation map (e.g., documents) were indications of the number of times a document had been CITED BY ANOTHER DOCUMENT.
        - I had thought that the proximity of two nodes were an indication of how frequently two documents had been co-cited.
        Based on these assumptions, I wrote in a ms. the following: "Density of 'links' connecting two scholars reflect the number of times a scholar has been co-cited by another scholar."
        I guess I am confused!
        Please reduce my confusion! :)
        • Ludo Waltman November 17th, 2018 9:18 am
          The proximity of two documents indeed is an indication of how frequently the documents have been co-cited. A link also represents the frequency of co-citation of two documents. It does not represent the number a times a document has been cited by another document. The expression 'co-cited by another scholar' should have been 'co-cited with another scholar'. I hope this offers some clarification!
          • Philip H. November 17th, 2018 11:17 am
            Therefore, 'denser links' between two scholars would indicate higher co-citation of their work by other scholars. Your explanation leads to 2 follow-up questions.
            1. If proximity of authors and the number of links between two authors BOTH reflect the degree of co-citation, then why are some authors with many links far away from each other on the map?
            2. Does any info provided in the map offer an indication of 'which scholars' are frequently co-citing other scholars?
            Thanks. Your prompt and clear feedback is really appreciated by all of us users of VOSviewer.
            • Ludo Waltman November 17th, 2018 12:30 pm
              1. There are two reasons for this. First, only two dimensions are available for the visualization. A high-dimensional data set therefore needs to be projected into two dimensions. This means that the data set cannot be represented in a perfectly accurate way. For this reason, strongly linked items are sometimes located far away from each other. Second, the distance between two items in fact indicates the strength of the relation between the items after normalizing for the fact that some items (e.g., authors with many publications) tend to have stronger relations than others (e.g., authors with just a few publications). The normalization is selected on the 'Analysis' tab in VOSviewer. The default normalization is the so-called association strength. It is also possible to turn off the normalization.
              2. No, this information is not provided. However, the map does show which authors are frequently co-cited with other authors. These are the authors with a high total link strength. They have a large size in the map, so they are prominently visible.
      • Nikhil November 17th, 2018 12:36 pm
        Thank you for your reply.
  • Pedro November 13th, 2018 6:22 pm
    Hey Waltman. How are you?
    I'm trying to use the VOSVIEWER for a survey paper. Nevertheless, I have a doubt about the Thesaurus file. Can you help me?
    So, as database papers, I send you this file from Scopus (it is a bit old and is not having problems as the recent files from Scopus database) . In this database, I have to replace some keywords, e.g. micro grids by microgrid; micro grid by microgrid, etc, as you can see in the file Thesaurus_1.
    On other hand, I have to standardize the name of an author: thorp, j by thorp, js, as you can see in the file thesaurus_authors.
    In both files, I think that I follow correctly your example of thesaurus files, however, when I try to generate the results, nothing happens. Can you point me where I'm wrong?
    Thank you in advance for your help, Mr. Waltman.
    I emailed you with my files.
    • Ludo Waltman November 13th, 2018 6:57 pm
      You need to use the Scopus format for author names (rather than the Web of Science format). This for instance means that 'thorp, js' should be 'thorp j.s.'. This should solve your problem. If it doesn't, then please let me know.
      • Pedro November 14th, 2018 1:22 am
        Thanks, Ludo!
        It is working now!!
  • Ana Paula Ermel November 13th, 2018 12:46 pm
    I having a problem successfully uploading a Scopus data file. When I downloaded the data file of the Scopus (CSV), and upload in Voswier, gives an error message of an incorrect number of columns in row 276.
    • Ludo Waltman November 13th, 2018 1:13 pm
      Unfortunately, there are currently problems with CSV files exported from Scopus. We are in contact with the Scopus team about this. They are investigating the issue. Hopefully, the problems will be fixed as soon as possible. As a workaround, you could consider using other data sources, such as Web of Science or Dimensions.
      • Ana Paula Ermel November 13th, 2018 1:49 pm
        Thanks, Ludo!
        I'll be waiting.
        • Ana Paula Ermel November 14th, 2018 12:56 am
          I just discovered that by unchecking the "Funding details" option when exporting the CSV file, VOSviewer runs it normally.
  • Oloiva Sousa November 11th, 2018 9:36 pm
  • Oloiva Sousa November 10th, 2018 8:38 pm
    i think that this application is excelent!
  • Philip November 10th, 2018 10:49 am
    One of my doctoral students is having a problem successfully uploading a scopus data file. Although she did not have a problem in the past, when she updated the data file in scopus and downloaded again, VV gives an error message of an incorrect number of columns in row 276. her old file is still ok.
    We've tried exporting a fresh csv file, renaming the file, deleting a record and saving under a new name etc, but no luck. Any suggestions?
    • Ludo Waltman November 10th, 2018 5:14 pm
      Could you please send your files to I will then take a look at your problem.
  • Angel Bravo October 31st, 2018 8:09 pm
    Hi, would you tell me where I can find the thesaurus_authors.txt archive please
    • Ludo Waltman November 1st, 2018 7:49 pm
      See the downloads available at
  • Philip Hallinger October 30th, 2018 1:47 am
    Hi: I understand that the overlay visualization in co-word analysis will reveal the 'research front' in the particular literature. I would assume that the research front for the current literature would be the last five or six years.
    1. Therefore, I don't understand why the default for a temporal overal in a review that extends from 1960 to 2018 only shows the research front for 2006-2014.
    2. Also when I try to adjust the years in the legend bar to be 2010 to 2018, all of the topics change to a darker color.
    What am I missing here?
    • Ludo Waltman October 30th, 2018 7:24 am
      The colors show the average publication year of the articles in which a particular term occurs. These averages typically are located somewhere in the middle of your time period. It is unlikely that a term occurs exclusively in early years or exclusively in very recent years. This is why there will be hardly any terms in your data that have for instance 1965 or 2016 as their average publication year. In the case of your data, most terms apparently have an average publication year between 2006 and 2014, and the color scale is automatically adjusted to this.
Build on Applepie CMS by Waltman Development