GraphXML — An XML-based graph description format
I. Herman, M. S. Marshall
Centre for Mathematics and Computer Sciences (CWI)
P.O. Box 94079, 1090 GB Amsterdam, The Netherlands
Email: {I.Herman, M.S.Marshall}@cwi.nl

ABSTRACT
GraphXML is a graph description language in XML that can be used as an inter-
change format for graph drawing and visualization packages. The generality and
rich features of XML make it possible to define an interchange format that not only
supports the pure, mathematical description of a graph, but also the needs of in-
formation visualization applications that use graph–based data structures.
Keywords and Phrases: information visualization, graph visualization, user inter-
faces, XML
1 Introduction
Most graph drawing systems could benefit from a textual description language to input
and, possibly, to output graphs in a human readable form. Ideally, such a description
language could also serve as a standardized format for data exchange between systems,
enabling information exchange. Defining such a description language is not, by itself, a
particularly difficult task: after all, a graph is simply a collection of nodes and edges.
Several formats have been proposed in the past such as GML[1] and WebDot’s DOT
format[2], and the formats used by Rigi[3], LEDA[4], and GDS[5]. None of these are
universally supported and they are usually bound to specific systems.
An important development of recent years, which also influences the choice of inter-
change formats, is the synergy between graph drawing techniques and information
visualization. Information visualization has become a well-known research area, with
important industrial and scientific applications. Graph drawing techniques play a
prominent role in information visualization because the data structures to be visualized
can often be described as graphs. The demands of information visualization pose new
challenges with respect to graph description formats. It is necessary to include features
in a graph description language that are not directly relevant to pure graph drawing. For
example, the graph description should be able to include application dependent data*,
either embedded in the description itself or externally referenced (e.g., network statis-
tics for a Web visualizer, genetic data for consensus trees as used in evolutionary re-
search, database references for the result of a database search, etc.). Furthermore, it
should be possible to describe composite structures such as nested hierarchies and

*
Although GML, for example, is a capable description language for graph drawing purposes and includes
provisions for extension, the mechanism for associating external data with a graph element is not well de-
fined.

2

clusters. Another less obvious feature is to support the description of the evolution of
graphs in time for applications such as animation.
As part of a larger project in graph visualization, we needed a graph description lan-
guage. Although, initially, we looked for an existing standard, like the ones cited
above, none could fully support the needs of information visualization. Consequently,
we decided to develop our own format with particular attention to the requirements of
information visualization. The result is GraphXML, a graph description language based
on XML. Description of this language is the subject of this paper. We hope that this
format will enter widespread use. In our view, this would be beneficial for the graph
visualization community.
2 Why XML?
XML is a language specification developed by the World Wide Web Consortium that
has received significant attention in the last few years†. The future evolution of the
Web, both in the traditional Internet domain and in mobile communications, is based
on XML. There are several reasons for choosing XML as the basis for a graph descrip-
tion format:
XML defines clear syntactic rules for specifying a “language” for a particular
application. An application-specific language is defined in a file called a Document
Type Definition (DTD). The end–user also has the option of adding extensions to
the language specified in the DTD.
XML is used by many different applications to define data formats including those
for databases, chemical compound definitions, e-commerce, mobile devices, and
schematic graphics. A graph interchange format based on XML has a greater
chance of being accepted by other application communities.
There are a number of XML–based specifications that are being defined by com-
munities, both within and outside the World Wide Web Consortium. GraphXML
can take advantage of some of these existing standards. See Section 7 for examples.
Software tools are emerging, which are either based on XML or work with XML.
For example, there are a number of both commercial and public domain XML edi-
tors. These tools can be very helpful in managing graph description files that are
based on GraphXML.
Several XML parser tools are freely available in Java, C, and C++. A full-featured
parser that provides error management and syntax checking can easily be generated
using these tools. The main task is to define the semantic interpretation specific to
the application (in our case, GraphXML).



†
There are hundreds of books on XML, so rather than pick a specific reference we refer to the original
specification[6], which is available on-line.