> -----Ursprungligt meddelande-----
> Frĺn: Social Network Researchers [mailto:[log in to unmask]]För Ulrik
> Ämne: Re: Get connected... Tool Integration Project
> On Fri, Apr 05, 2002 at 03:48:21AM +0200, Kósa Dávid wrote:
> > 2. Examine the underlying data structures and create a generic
> and flexible
> > structure that could be an interchange format for all software
> tool. This
> > generic network data structure, that would be supported by SNA software
> > packages, would elliminate the need of conversions.
> The proposal for a modern XML-based network data format called GraphML has
> recently been published. It is capable of handling graphs, hypergraphs,
> hierarchical graphs, and collections of graphs. See
> U. Brandes, M. Eiglsperger, I. Herman, M. Himsolt, and M. S. Marshall:
> GraphML Progress Report: Structural Layer Proposal.
> Proc. 9th Intl. Symp. Graph Drawing (GD '01), LNCS 2265, pp. 501-512.
> © Springer-Verlag, 2002.
> Extensions for light-weight parsers and graph attributes (valued networks)
> will be released within the next two month and we are working on an
> extension for optional inclusion of graphics information. Several major
> graph drawing tools have made commitments to support the format, and
> Java SDK 1.4 has built-in support for XML.
The GraphML XML-type language seems like a promising way to get a unified
format for describing graphs of different kinds. It will probably be very
handy for SNA and other fields concerned with graph drawings.
Nevertheless, I doubt if GraphML will become the standard language for
social network analysis in general - to store a sociomatrix in GraphML is a
lot more memory-consuming and parsing-demanding than storing the same data
in a tab-separated file such as a csv-file (comma-separated values), even
for non-valued, dichtonomous networks. Also, most spreadsheet programs up to
date are familiar with importing and exporting csv-files and I doubt that
future versions of Excel etc will support the importing of GraphML-files.
(Such imports would nevertheless require that Excel then parses the
GraphML-files and converts them - I very much doubt that GraphML will be the
standard format when storing spreadsheets!)
What Kósa Dávid was asking for was perhaps a standard format for storing SNA
data. What about csv-files? Such files can easily be converted into graphs
too - below is a link to a spring-embedder written in Java that I have
developed recently which imports csv-files taken directly from my server,
and it's quite quick when converting the csv-file into a graph structure
(consisting of nodes- and edges-objects in the Java object-oriented style).
This currently only supports one-mode networks but it would be easy to
include by having separate csv-files for each different edge-types. It
doesn't however support actors' attributes - and I suppose that it would
require something more structural than a csv-file to include that (although
it can be supported if there was a standard agreement on how actors'
attributes are to be stored - once again, as with multi-mode networks,
perhaps multiple csv-files can be a solution?)
- - -
Carl nordlund, PhD student
[log in to unmask] <mailto:[log in to unmask]>
Human Ecology Division, Lund University, Sweden
> The project homepage www.graphdrawing.org/graphml is currently being
> redesigned, but will have updated in formation in the future. A public
> discussion list will open soon.
> A one-serves-all unified tool combining the functionality of those
> existing seems undesirable to me. IMHO, it is necessary to have tools
> at different levels of sophistication for different groups of users.
> A large open-source project like those typical for Linux environments
> is simply unrealistic. The closest you might get in the context of
> social networks is Carter's stuff at
> Dr. Ulrik Brandes
> University of Konstanz | +49 7531 88 4263 (phone)
> Department of Computer & | +49 7531 88 3577 (fax)
> Information Science, Box D 188 | [log in to unmask]
> 78457 Konstanz, Germany | http://www.inf.uni-konstanz.de/~brandes/