Full program - speakers, titles and abstracts - now available.

-----------------------------------------------------------
International Workshop on Social Space and Geographic Space
(SGS'07)

in conjunction with the
Conference on Spatial Information Theory (COSIT'07)
(http://www.cosit.info/)

19 September 2007, Melbourne, Australia
-----------------------------------------------------------
 
AIMS AND SCOPE
==============
Social agents are embedded in both social structures and in geographical space. The combination of social and geographic space has often been neglected. With few exceptions, social network theory ignores geographic space, and artificial intelligence studies often assume local societies without sophisticated conceptualisations of social networks. Yet the multiple embeddedness of actors in both physical and social space has important implications for understanding social behaviour.  In many related research areas, there is a growing recognition that associations between social structure and geographical nearness may affect social systems and social behaviours.

Research on the associations between social and geographic space occurs in disconnected scientific communities, including human geography, social network theory, and geographic agent-based simulation. The Workshop on Social Space and Geographic Space aims to bring the social and the spatial disciplines together, to discover joint foundations in social and geographical theory, and to integrate approaches for modelling spatial context and social behaviour.

The workshop is organized as an ARCNSISS activity, but open to all researchers. Participation in the following Conference on Spatial Information Theory is optional and facilitated.

PROGRAM
=======

MORNING SESSION

Keynote:
Carter Butts (UCI, USA): Social Networks in the Physical World

Though often viewed as aspatial abstractions, social networks represent relationships among actors in the physical world.  As such, these relationships operate under spatial and temporal constraints that have substantial implications for the structures they constitute.  More profoundly, the joint consideration of space and structure gives rise to a "social geography" whose properties are not reducible to those of either element in isolation.  Here, I introduce a number of basic results in the measurement and analysis of spatially embedded social networks.  These include the use of spatial interaction functions to represent spatial dependence, modeling of spatially embedded networks via curved exponential families, and the extrapolative simulation of spatial macrostructure.  A number of sample applications will be shown, ranging from group interactions to interorganizational collaboration networks.

Galina Daraganova and Pip Pattison (Melbourne, AU): Social Networks and Space

This paper examines external spatial constraints on network processes, generalising the exponential random graph class of network models. We develop a hierarchical set of nested models for social networks embedded spatially, in which the inverse square of the Euclidian distance between them is a dyadic covariate. Using a simulation approach based on the Metropolis algorithm, we explore spatial embeddings of graphs with different global features of network structure (random graphs and graphs with small–world properties). Simulation results demonstrate the way in which individuals’ geographical locations and endogenous network clustering processes combine to structure patterns in social networks.

Scott Baum (Griffith U, AU), Anthea Bill (Newcastle, AU) and William Mitchell (Newcastle, AU): Labour Market Disadvantage in Metropolitan Regions

There has been growing awareness that in order to understand issues of social disadvantage it is necessary to undertake an appraisal that includes reference to the impact of individuals and the local contexts they operate in.  By focusing on the interactions between local labour market regions and individual characteristics this paper presents an analysis of unemployment risk that moves beyond studies that simply focus on the individual aspects of unemployment or the broad patterns that exist across spatial aggregations. The paper finds that significant weight should be given to explanations of unemployment that move beyond a simple focus on individual employability by considering the impacts of broader contextual factors that may operate net of individual factors.

Bill Hillier (UCL, UK): What Do We Have to Add to a Social Network to Get a Society? Answer: Something like What We Have to Add to a Spatial Network to Get a City

Recent years have seen great advances in social network analysis. Yet, with a few exceptions, the field of network analysis remains remote from social theory. As a result, much social network research, while technically accomplished and theoretically suggestive, is essentially descriptive. How then can social networks be linked to social theory? Here we pose the question in its simplest form: what must we add to a social network to get a society? We begin by showing that one reason for the disconnection between network theory and society theory is that because it exists in space-time, the concept of social network raises the issue of space in a way that is problematical for social theory. Here we turn the problem on its head and make the problem of space in social network theory explicit by proposing a surprising analogy with the question: what do you have to add to an urban space network to get a city. We show first that by treating a city as a naïve spatial network in the first instance and allowing it to acquire two formal properties we call reflexivity and nonlocality, both mediated through a mechanism we call description retrieval, we can build a picture of the dynamics processes by which collections of the buildings become living cities. We then show that by describing societies initially as social networks in space-time and adding similar properties, we can construct a plausible ontology of a simple human society.

AFTERNOON SESSION

Keynote:
David O’Sullivan (Auckland, NZ): The production of space in social simulation models

In this paper I will present work which, over a number of years, has sought to include richly structured representations of spatial relations in dynamic simulation models.  A simple agent-based model of segregation will be described that progressively includes the capacity for its spatial structure to change in a variety of ways in response to the spatial distribution of the agent population.  In this model, the actions of agents collectively produce reconfigurations of the space, which in turn act back on the behaviour of agents.  Prospects for the explicit incorporation of social networks in this kind of model will be considered.  Implications of this model for analytical approaches will also be examined.

Sabine Timpf (Würzburg, Germany): Simulating Place Selection in Urban Public Parks

How do people choose their place when walking into an urban public park? This question is important in the context of socially sustainable recreational spaces and thus interesting to park managers and designers. Our aim is to systematically investigate park management and design options using analyses and simulations of the behavior of park visitors. The first step towards this goal is to model place selection in public parks. How can we model this highly individualized and dynamic process? One theory asserts that park users perceive affordances of the park furniture and choose a place according to their needs. Needs are largely influenced by the intended activities but also by expected social interactions. In our multi-agent simulation we model the notion of affordances of park furniture with respect to (a small number of) goal activities. Preliminary simulation results indicate that affordances and activities alone are not sufficient to explain place selection. Currently we are implementing the notions of personal and social space as well as affordances of park users, thus introducing a social dimension into the model.

Daniel Loiterton and Ian Bishop (Melbourne, AU): Simulating Visitor Behaviour: Algorithms for a Recreational Agent Model

Autonomous agent models are becoming increasingly common in studies of visitor behaviour in recreational landscapes. As we begin to understand the complex relationships between people and their environment, we realize that simple empirical spatial models are insufficient for predicting human landscape impacts. The potential for accurate visitor simulation does exist however, via the integration of spatial information and intelligent, mobile agents. Such a system is currently being developed at the University of Melbourne, Australia, which aims to replicate visitor movement patterns in the Royal Botanic Gardens (RBG) Melbourne. The Intelligent Recreational Agent Simulator (iRAS) uses autonomous reactive and proactive agents to produce a range of behaviours observed within the real environment. Each agent can perform specific location-based actions, motivated by their unique set of feelings and goals, or they can take a leisurely stroll through the path network, reacting to the physical and visual qualities of the landscape, as well as the presence of other agents. When hundreds of these individuals are generated using input data obtained from field-based surveys and behavioural experiments, the resulting collective movement patterns correlate exceptionally well with observed reality. This paper will introduce the approaches used to create, parameterize, calibrate, and validate the iRAS agent system, but will focus on specific algorithms used to model the required agent behaviours.

Steven Hite (Brigham Young U, USA), Julie Hite (Brigham Young U, USA), Christopher Mugimu (Makerere U, Uganda), and Joshua Rew (Florida State U, USA): Geographic and Social Space: A Statistical Analysis of Euclidean, Actual, and Least-Cost Distance and Network Ties of Headteachers in Uganda

Using geographic and social network data collected in the field, this study specifically explores the potential statistical difference between three distance measures (Euclidean, Actual, and Least-Cost). The study uses distance measured three ways between dyadic ties in a network of secondary school headteachers (n=58) in Mukono Uganda. While the mathematical distinction between these three measures is self-evident, their potential for differential use in statistical analysis is the focus of this study.  Thus, the study explores three questions regarding the three distance measures: 1- are they statistically different (using Pearson correlation and Cronbach’s Alpha), 2- are they different in identifying the existence of network ties (using t-tests), and 3- are they different in predicting network ties (using Logistic regression)?  Extending from the findings of the statistical tests for differences, the implications of four factors of data cost for the three measures are then explored.

Pragya Agarwal (UCL, UK) and Stephan Winter (Melbourne, AU): Place as a Constituent for Social Networks

Social network analysis has traditionally ignored the role of place and geographic space in forming social bonds and networks. Here we introduce a richer model by introducing the notion of co-placement as a factor for social bonds. A hierarchical categorization of co-placement and the inferred strength of relations at different scales and granularity are discussed. This research shows that the geography of places is instrumental in determining the strength of social bonds, which is studied using the case of academic collaborations in form of published papers between researchers in the GIScience and COSIT conference communities.

PROGRAM COMMITTEE
=================
* Pragya Agarwal, University College London, UK
* Carter Butts, University of California Irvine, USA
* Christophe Claramunt, Naval Academy Research Institite,  F
* Matt Duckham, University of Melbourne, AU
* Bruce Edmonds, Manchester Metropolitan University, UK
* Joachim Gudmundsson, NICTA, AU
* Patrick Laube, University of Auckland, NZ
* Pip Pattison, University of Melbourne, AU
* David O'Sullivan, University of Auckland, NZ
* Paul Torrens, Arizona State University, USA

ORGANIZERS
===========
Stephan Winter
Dept. of Geomatics, The University of Melbourne, Australia
email: This email address is being protected from spam bots, you need Javascript enabled to view it
phone: +61 3 83447875
url:   http://www.geom.unimelb.edu.au/winter/

Garry Robins
Dept. of Psychology, The University of Melbourne, Australia
email: This email address is being protected from spam bots, you need Javascript enabled to view it
phone: +61 3 83444454
url:   http://www.psych.unimelb.edu.au/people/staff/RobinsG.html

SPONSORSHIP
===========
The workshop is sponsored by the Australian Research Council Research Network on Spatially Integrated Social Science (ARCRNSISS), but open to all researchers.