Source
Pajek was developed by the University of Ljubljana, and can be downloaded from its web site
Cost
Pajek is freely downloadable, and appears to have no licence conditions attached. When used for analysis, one or more of the following references should be cited:- V. Batagelj, A. Mrvar: Pajek - Program for Large Network Analysis. Home page
http://vlado.fmf.uni-lj.si/pub/networks/pajek/. - V. Batagelj, A. Mrvar: Pajek - Program for Large Network Analysis. Connections 21(1998)2, 47-57.
- V. Batagelj, A. Mrvar: . Pajek - analysis and visualisation of large networks. To appear (September 2003) in Graph Drawing Software book, edited by M. Jünger and P. Mutzel, in the Springer series Mathematics and Visualization.
Help files
The home page has links to the official documentation. There is also a Manual.
Core Capabilities
Pajek is a package for the analysis of large networks. It can handle very large networks, with thousands of nodes and tens of thousands of links. Both single node and bipartite networks can be analyzed. A wide range of published network analysis algorithms have been implemented in Pajek, including such staples as network diameter, cluster coefficient and density, and most run very rapidly.Pajek uses a strictly defined, text-only file format for reading and writing networks. Although converting data sets to the Pajek .net format can be a tedious business, the ascii files are human readable, allowing easy inpection of the raw data.
The program also provides excellent network graphics, allowing networks to be displayed in a variety of layouts, once again following published algorithms. Nodes can be colour coded according to different chatacteristics such as component membership or degree, and nodes and links may be labelled or unlabelled. There is full control over the size, colout and position of nodes, and finished network graphics can be exported in a number of common image formats.
Usefulness
Pajek is particularly useful for the analysis of very large networks, since most of the algorithms run quickly and scale well. It is particularly useful for exploratory network analysis, since it includes many algorithms which cen be applied to a network in sequence.
Limitations/Weaknesses
The major limitatation is the lack of access to the underlying code. Either source code access or some form of scripting capability would be useful for researchers interested in developing new algorithms, rather than just applying existing ones.Pajek's documentation is something of a weakness. Although there is an onlune manual and tutorial, neither is particularly informative. It is particularly difficult to find details of the algorithms implemented, and where the original algorithm was published - a serious issue for researchers.
A feature which would be nice to have, but probably doesn't count as a serious weakness, is the ability to read file formats other than the Pajek .net file. However, given that there is no widely agreed standard format for network description, this is probably asking a bit much.
Screen Shots
This is the main Pajek interface:
An example network graphic:
Other Information
There is a mailing list for Pajek, and the authors are very active in network layout competitions at various conferences. Links are available on the main Pajek web page.
Administrivia
This page prepared by Jennifer Hallinan, Institute for Molecular Bioscence and School of ITEE, The University of Queensland.Last edited 26/07/2003.