University of Konstanz, Computer Science

Communication and Exploration:
Issues in the Visualization of Social Networks

Sunbelt XVIII and 5th European International Social Networks Conference are held jointly in Sitges, Spain, May 28-31, 1998. A session on the visual presentation of social networks is organized by an interdisciplinary team of researchers from the University of Konstanz working on the visualization of social networks.

Session overview

Additional Information

Abstracts

Cancelled talk:

Helen Purchase (University of Queensland)

Graph Layout Aesthetics and their Application to the Visualisation of Social Networks

Abstract:The development of effective software environments intended for the visualisation of social networks has not progressed as rapidly as in other areas. Freeman has investigated the potential application of molecular visualisation tools and some domain-independent automatic layout algorithms to social networks, but no specific conclusions have been reached as to the most effective manner of depicting social networks. McGrath et al have considered the problem experimentally: while their experiment clearly demonstrates that the form of the layout of a network affects the perception of clusters, specific guidelines as to the most effective manner of laying out a social network are yet to be proposed.

This paper will address the problem of the automatic two-dimensional layout of social networks by considering the particular aesthetics on which many layout algorithms are based (eg. minimising the number of crosses and bends, maximising the display of symmetric structures). I will propose that for the display of a social network to be effective, particular aesthetics which highlight the relevant information need form the basis of the design of the automatic layout algorithm that produces it. The measurement of these aesthetics, and experiments which may make use of them in the context of social network visualisation will be described.

Cathleen McGrath (Carnegie Mellon University) and James Blythe (University of Southern California)

An Empirical Investigation of Users' Inferences from Graph Layouts

Abstract:The visual display of network data is an important mechanism for uncovering and communicating social network phenomena. Previous work has shown that the layout of social network graphs can influence viewers' perception of network properties. In this study we move toward understanding how display influences perception by using an interactive experimental design. We are able to track and time viewers' assignment of nodes to groups in order to capture how viewers interact with the graph. We use time to complete group assignment as a measure of the clarity of the graph. We also examine the order of assignment of nodes to groups and the number of times nodes are re-assigned to understand how viewers' attention focuses on different sections of the graph. Understanding how viewers interact with graphs will contribute to static and dynamic display of social network data.

Lothar Krempel (University of Cologne) and Thomas Plümper (University of Konstanz)

The Dynamics of International Economic Integration:

Locations of Change in International Trade identified with Multivariate Estimates of a Series of Gravity Models and Visualisations of their Error Terms

Abstract: In recent research it has become common to argue that the increasing integration of world markets has drastically changed the conditions of economic policy-making in the advanced industrialized countries. Moreover, it has become widely accepted that globalization is a phenomenon that covers all countries and world regions.

Jeffrey Frieden and Ronald Rogowski (1996), argue for example that the tendency towards a closer integration stems from a relative easing of international economic transactions compared to national business activity. This exogenous decrease in the costs of international economic exchange is seen as a result of technological change and political liberalization.

An alternative minority view understands global economic processes as the result of global economic imbalances. The increasing budget deficit of the industrialized states and the growing differences in national saving and investment figures serves as an engine of international capital flows: due to the accounting identities between capital flows and trade in goods and services the negative capital flows have to be balanced by international trade (Krugman 1990; Krugman 1996).

Krugman expects bilateral economic interactions to change most between countries, with large imbalances of national savings and national investment. His theory does not assume that change in economic transactional stems from a global integration.

Countries characterized by a trade deficit are often net importers of capital , whereas countries with a savings surplus export capital which contributes to their trade surplus. The most notable country with a national saving deficit is the USA, which is also true for the most of the rapidly developing countries of South East Asia. Japan on the other hand is a prominent example for a country with a high surplus.

Concerning the location of growth we should therefore expect most change to occur within the pacific area, while the transatlantic and European economic relations should be fairly stable.

We compare both competing views with the help of a series gravity models, which we estimate by multiple regression. A visual inspection of the trade estimates (and their error terms) with NetVis, a tool developed for the visualization of interaction networks - which also works well for international trade flows - allows to identify regions of strong and weak growth on the basis of a given model specification. The suitability and sufficiency of various explanations is thus evaluated on a thorough analysis of the residuals.

The above outlined procedure allows to formulate a model of global economic integration, which takes the explanations of the above cited theories into account.

Ulrik Brandes, Patrick Kenis, Volker Schneider, and Dorothea Wagner (University of Konstanz)

Towards Effective Network Visualization

Abstract: Visualization is an important aspect of both exploration and communication of all kinds of data. Graphical displays of social networks are particularly attractive, since they enable researchers to display in a compact way the position of individual actors in a network, how they are related to each other, and what the overall structure looks like. Sociograms were early companions of social network analysis, but have received surprisingly little attention during the following decades. Only in the last few years, easy accessability of quality computing and graphic equipment has revived a now rapidly growing interest.

It is argued that, in order to produce effective visualizations of social networks, it is necessary to be concious about three major aspects:

• the substance to be communicated/explored,
• a design which appropriately maps the substance to a graphical presentation, maps the substance to a graphical presentation,
• and an algorithm that realizes the design sufficiently.

The ultimate goal is to devise means for assessing the effectiveness of social network visualizations, and improving automatic drawing procedures accordingly. As a first step, the paper presents a formal framework based on semiotic principles, which is intended to lay the ground for such developments.

Vladimir Batagelj and Andre Mrvar (University of Ljubljana)

Drawing Genealogies

Abstract: There exists a huge amount of (often very large) genealogies already in computerized form and accessible on the Internet: http://www.gentree.com/ For their interchange a special format, GEDCOM, is used.

From these data two types of graphs (Ore's graphs and p-graphs) can be derived. In the paper we present some tools for analysis (genealogical distance between persons, searching for marriage patterns, statistics, ...) and nice drawing of such graphs. They are implemented in Pajek - a program package for analysis of large networks.

Noshir S. Contractor, Mike Chan, and Dan Zink (University of Illinois at Urbana Champaign)

IKNOW: A Tool to Visualize, Assist, and Study the Creation, Maintenance, and Dissolution of Knowledge Networks

Abstract:The introduction of new communication and information technologies in organizations have primarily been used to create new channels of communication and/or reduce the cost of communication among members in the workplace. Ironically, the pervasiveness of electronic communication media in virtual organizations and communities makes it increasingly difficult for individuals to discern social structures. Consequently, individuals have significant problems accurately determining Who knows who? and Who knows who knows who? Fortunately, information technologies that are responsible for triggering this problem can also be used to overcome these obstacles. Because information transacted over electronic media such as the Web can be stored in digital form, a new generation of software called collaborative filters or communityware (Contractor, O'Keefe, & Jones, 1997; Kautz, Selman, & Shah, 1997) can be used to make visible the organization's virtual social structure. One such tool, IKNOW (Inquiring Knowledge Networks On the Web), has been designed by a team of UIUC researchers to assist individuals to search the organization's databases to automatically answer questions about the organization's knowledge network, that is, Who knows what? as well as questions about the organization's cognitive knowledge networks, that is, Who knows who knows what? within the organization. Unlike traditional web search engines that help an individual search for content on the web, tools such as IKNOW search for, and visualize, content and contacts (direct and indirect).

Carter T. Butts (Carnegie Mellon University) and Kathleen Carley (Carnegie Mellon University)

Visualization for Comparison of Unlabeled Structures Using Existing Technologies

Abstract: Comprehension and communication of social network data is greatly facilitated by the use of visual display techniques. Such techniques, however, become problematic when one seeks to directly compare disparate social structures; this is even more true of social structures which do not possess meaningful theoretical labels. In principle, it is not possibleto visualize such unlabeled structures at all, since embedding them in a display space constitutes a form of tacit labeling. Using previous work on canonical labeling algorithms, however, we establish a simple approach which permits the representation of even unlabeled structures, and which facilitates visualization for purposes of structural comparison. This approach is illustrated using Krackplot, a well-known and accessible network visualization application. Techniques for extending this method to partially labeled structures, and for using this technique in conjunction with cluster analysis of social structures, are also discussed.

Greg Stephens (University of New South Wales)

Network Visualization using VRML - A Prototype

Abstract: One of the problems with visualization techniques currently available is that they require the viewer to be at a distance from the network looking in . The network so described can usually be rotated on either axis around the geometric centre of the network. The application of the Virtual Reality Modelling Language (VRML) removes both of these restrictions. The viewer can rotate the network about any node, consequently being able to view the network from that node's perspective. The zoom-in, zoom-out feature of VRML enables the user to enter the network perimeter and so focus in on groupings that are observed while still being able to rotate the network on either axis.

This prototype, which will be demonstrated, was designed using output from UCINET which was then configured into VRML via a Microsoft Visual Basic application. The VRML application used to view the network is Cosmo Player 2. The output from UCINET is used to determine the location of nodes in three dimensional space, as well as, provide input into the algorithm which determines the strength of the bond that connects nodes. The difficult part of this application was the specification of an algorithm to place the bonds in three dimensional space between two nodes.