Pajek (Slovenian: spider) is a software for large network analysis which is free for non-commercial use and can be downloaded from its authors' site. Once downloaded, double-click the file to install Pajek on your computer. Usually, answering yes to all installation questions results in a working program.
After installation, when Pajek is called by a double click, Pajek presents its Window which has a set of menus lined up on the first row: and so on until . The Pajek Window is identified by the title on the top left window next to the Pajek icon (a spider, of course). The Pajek Window also has a set of objects and icons on the first column downward, namely , and at the bottom of the column. Each of these objects has three icons immediately below, namely the yellow Read icon, the white-black Save icon and the white-yellow Edit icons. If you let your mouse cursor rest on one of these yellow Read icons for a second, it will show a description of the icon on a yellow background, such as Read Network or Read Partition.
In this introduction we will be working only with a handful of
menu choices, namely
and
.
Please note the difference between singular Net or Partition and plural Nets or Partitions. They present different menu choices depending on the number of objects to manipulate. We will also be working only on Network and Partition with their icons Read in yellow and Save in white-black. Regarding the last icon, Edit Network, editing of objects in Pajek is not recommended because it is much easier to do editing with an ASCII editor or word processor you are familiar with.
To follow the next discussion, you should have downloaded and installed Pajek following the above description. You should also preferably have prepared a network file or download one network file from the many repositories on the Internet such as de Nooy et al. 2005 data.
Double clicking the Pajek icon opens the Window like above. To open a network, click on the Read Network icon which will bring up the window below. Again, you identify the window by its title on the top left. Double click on the icon (on the left) of hawthorne-friend.net to open it. Pajek responds by reporting that it has read the network in a flash (Time spent: 00:00:00). In fact it took less than a second. This kind of information might be useful later when you work on large datasets such as over 3 million patents granted by the United States Patent and Trademark Office which is available from Vlado's site.
-----------------------------------------------------
Reading Network --- C:\temp\hawthorne-friend.net
-----------------------------------------------------
Working...
29 lines read.
Time spent: 00:00:00
Close the Window by clicking on the cross (X) at the top right of the window. Again, you identify the report window by its title on the top left which says Report. After a while you may need a short cut to quickly close this Report Window, do so by pressing both Alt+F4 while it is active, that is when its top bar is coloured blue.
At this stage, verify that you have the correct network by making
sure that c:\temp\hawthorne-friend.net is the working network on your Window. More precisely, hawthorne-friend.net is shown in the network area. Additionally it says that the network has 14 vertices.
During an analysis session, you might be working with more than
one networks, say hawthorne-friend.net and malaysia.net. You can open more networks following the procedure above. In fact, you might want to try opening a few of the networks in de Nooy et al. 2005 data. Pajek may also need to create a few networks during the course of analysis, for instance during the analysis of network dynamics. In that respect, Pajek has the ability to provide many networks, many partitions,
many permutations objects at your disposal at any instance of
time. To keep track of which network you are currently working on
make sure you select the intended network. You can list what are
the available networks by clicking on the triangle on the far right
on the network area. You select a network by moving the cursor up
and down the list and clicking on a selected network. A similar
principle of operations is effective in selecting different
objects such as partition as the working object.
To visualise the sociogram of friendship network in the Hawthorne study, you draw the network by clicking on the menu on the first row of window, then move down to select choice. From now on, a shorthand for this sequence is .
It should open a Draw window like below with the 14 friends and the friendship relations between them drawn. The title on the top-left says the name of the network drawn and there are 14 vertices in all. After a while you might find the shortcut for drawing operation handy, just use Control+G.
To have a nice looking sociogram, on the Draw window (i.e. the one with the network name as the title of the window) choose . See figure on the left below. The result is a nice layout of sociogram of friendship network. See figure on the right below. Further information about the principles of laying out this sociogram can be found on Kamada and Kawai (1989) and de Nooy et al (forthcoming). There are other alternatives layout algorithms available in Pajek, like the Fruchterman and Rheingold (1991) algorithm, which are also discussed by de Nooy et al (forthcoming) in relation to visualising social networks.
|
|
Once you have your sociogram, you can save or export it to be
included later in your documents using e.g. Microsoft® Word or your favourite TeX™ editor. Pajek can export drawing into many formats,
including Encapsulated PostScript (eps), hypertext mark-up
language (html) with Scalable Vector Graphics (svg) and Windows
bitmap format (bmp). For example, Microsoft® Word 2002 and later,
can handle encapsulated postscript files quite easily, so eps
format is a safe option. To export it, in Draw window choose
and Pajek will ask for a file name in a dialog
box. Give it a name, e.g. hawthorne.eps and then save it. This file can then be included in your editor or word processing package as usual. Before saving your sociogram, however, you might play around with positioning vertices by dragging them to new positions with your mouse cursor in order to achieve a result pleasing to you.
 |
 |
| Back to Introduction to Social Network Analysis | Back to top |
© G Tampubolon - 17 December 2004