28

I was wondering if there already exists a software/service that allows to build a tree of references/citing articles starting from a selected paper X. For example paper X has a list of references, call it bibliography[X]. This list of references is usually presented in the journal's webpage and, in fortunate cases, the list is also accessible by cross-reference. The idea is then to go to all the papers Y in bibliography[X] and repeat the process for the papers in each bibliography[Y] up to some depth.

On the other hand, we could also look at the citing articles of X, which is also displayed in the journal's webpage as well as in other resources, e.g. Google Scholar.

In this way the tree of references/citing articles of X up to some depth can be obtained and plotted in a nice graphical framework and perhaps display relevant information of the vertices of the tree like title, doi, etc. In my opinion this would be a very nice and efficient way of discovering new papers related to some paper, e.g. a seminal paper (citing articles tree) or to look into the history of a subject (references tree).

I have been looking for something like that for some time but so far nothing. I think Microsoft Academic Search would have the potential to do this by using the same framework they use for the Co-author path. Also there is this project for JabRef but it is not what I'm looking for.

Edit: Something like this would be perfect but unfortunately it is not available to the public.

  • 1
    I think it will only be useful when you add other filters, e.g. specific keywords or minimal number of citations. Otherwise it will be too large to display reasonably: e.g. 30 references per paper makes 900 nodes at depth 2. – silvado Oct 11 '13 at 7:27
  • 2
    Web of Science has this, but it's slow and ugly. – David Ketcheson Oct 11 '13 at 10:56
  • 1
    I totally agree with David. And I think the reason it's nearly useless is basically what silvado said; partly because of its UI, but mostly because, without some sort of intelligent filter, the tree becomes too huge to be of practical use at an exponential rate. – Yuichiro Fujiwara Oct 11 '13 at 22:44
  • 2
    @YuichiroFujiwara I don't think it is useless even without filter. See for example autodeskresearch.com/projects/citeology – Prastt Oct 12 '13 at 18:11
  • 1
    @Barefeg If I'm not mistaken, Citeology you linked to used only 3,502 papers. Accoding to Wikipedia, Web of Science now indexes around 65,000,000 items per year and covers data from the year 1900. I tend to think that your tree used a pretty strong filter and only proves my point. – Yuichiro Fujiwara Oct 12 '13 at 19:05
6

If you're serious about such a project, I think what you need to do is recognize that this task naturally separates into two parts. It will be very easy to find a good, general purpose and freely available program that suits your taste for visualizing/analyzing/discovering networks in general. I think cytoscape is pretty good, although I've only played around with it a little and many others exist.

Now, the second (well, first) part is to fetch and prepare the input data for the citation relationships of interest to you. You need to be very specific about exactly where you want to get this from and for what subset of articles, since it may not be possible (due to license restrictions) or technically feasible (as noted, the size of the network grows exponentially).

2

This is a tool if you use Zotero to manage your references and use the Related feature to link references together: http://people.ucsc.edu/~cmbyrd/zotnet.php.

1

Web of Science has a similar tool called Citation Map. but it is not free even on many public university campuses. I wonder if there are free tools/scripts who do https://youtu.be/Qr8wQnTHMMg

1

One way to do this is to use VOSviewer. It will not fetch the citation information for you, but you can download the results of a Web of Science search into a (series of) file that contain so called records with the cited references of each article in the search.

VOSviewer can build a citation graph from such files, and has a bunch of features to color the graph based on other publication info (publication date, number of citations, ...). The graph nodes may either represent publication, authors or keywords, and the vertices may represent citation, keyword co-occurrence, co-authorship, ...

Still, I wish there would be a better option, for the features of VOSviewer remain limited and the graph layout is not so easy to control.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.