Stephen J. Carver, D. Ian Heywood & Steven J. Tomlinson

Position paper for I-17

Dr Stephen J. Carver (steve@geography.leeds.ac.uk)
Dept of Geography
University of Leeds, Leeds, England

Dr Ian Heywood (i.heywood@mmu.ac.uk)
Dept of Environmental & Geographical Sciences
The Manchester Metropolitan University, Manchester, England

Steven J. Tomlinson (s.tomlinson@mmu.ac.uk)
Dept of Environmental & Geographical Sciences
The Manchester Metropolitan University, Manchester, England

This is a position paper for Initiative 17: Collaborative Spatial Decision-Making. We have been working on the use of computers as tools for spatial decision making in the following areas.

  1. The development of Idea Generation Systems (IGS) which utilise the spatial exploration tools embedded within Geographical Information Systems (GIS) to enable a greater understanding of the use of spatial and non spatial information for decision making.

  2. The development of prototypes which facilitate the use of meta spatial models to integrate data from existing spatial models(GIS and non GIS based). This allows a more detailed understanding of a problem domain to be constructed than individual models can supply and provides an overview of the consensus and conflict caused by the integration of the models.

  3. Using the above ideas we have developed software prototypes for a variety of real world case studies in participatory decision making. For example, the "House Hunting Game" which uses Idrisi and Toolbook.

1.0 Introduction

The development of computer systems which facilitate understanding and raise awareness of environmental problems are important to help us manage the environment in which we live. This type of computer system which utilises spatial and non spatial data, will become increasingly important as more information sources become available along with the need to understand its impact on people, projects and the environment - this can be termed computer based spatial decision making.

The use of computers to explicitly assist with spatial decision making is a relatively new occurrence. GIS are not complete decision making systems because they lack many of the requirements for decision making such as good user interfaces, modelling capabilities and so on (Densham, 1991, Heywood & Carver, 1994, Tomlinson, 1994). As well as the facilities that the GIS lack, they are not easy to use by non GIS specialists. Computer based spatial decision making should not be solely in the domain of the technical scientist who is adept at manipulating GIS and other software for research goals but should also be available for the less technologically proficient who makes the decisions or on whom decisions impact i.e. the policy or decision maker and the general public.

Through the design of more appropriate computer systems which address the needs of the users, such as managers and decision makers, the use of technology for spatial decision making can be taken from the technical domain into the general domain. This will empower people to utilise the technology to make decisions and better understand their environment.

2.0 Issues in Collaborative Spatial Decision Making (CSDM)

Prior to considering the uses and problems of CSDM it is first necessary to determine who the users of such systems are because this will impact upon how the systems should be developed and their eventual success. In the information systems literature, the integration of the potential users has been considered as a necessary part of the development of successful systems (Szajna & Scamell 1993, Lawrence & Low 1993, Wetherbe 1991, Joshi 1991). This is more so in the case of CSDM because the user is not a single identifiable person but potentially a group of people with a large "split personality". Thus, any system should be very easy to comprehend and use and be able to view problems from multiple perspectives which are dependent on the decision maker and problem domain. CSDM systems should be used as a medium of communication which enable the understanding and analysis to take place. Using a GIS, solely for such CSDM analysis is akin to looking through a fogged window.

Through techniques of software linking, such as Object Linking and Embedding(OLE), Dynamic Data Exchange(DDE) and loosely/tightly coupled software applications, it is possible to construct working environments which aid spatial decision making. For example, we have linked Asymetrix Toolbook, Visual Basic and Idrisi(DOS). Such linkages form the core of supra applications which provide the decision maker with the necessary tools from disparate software applications in a single environment providing a more information rich environment. Systems constructed using such techniques are the " House Hunting Game" and the "Kirklees System" which are described later. The users, of our example systems, have been the general public and policy/decision makers at city council level.

The development of CSDM systems requires a fundamental change in our approach to information management and use. If we are to use systems which enable multiparticipatory decision making, then these system must be easy to use and adaptable to various situations to enable the creative process of group stimuli and interaction to be transported into the CSDM system and give appropriate levels of feedback to the user(s). The CSDM must become part of the group and able to be used by all members to see what is happening and have control over the CSDM, this could be through a single user or multiple user interface systems.

One potential approach is that of the Idea Generation System(IGS). IGS is concerned with the development of CSDM systems to enable the decision maker to visualise their information in a dynamic environment which supports participatory decision making. Examples such as Sim City, integrate a variety of models and visually demonstrate change to the user. Environments which permit the decision maker to see what will happen given x or y and are then able to respond to such visual stimuli enable better decisions to be made through an awareness of not only the initial problem domain but of the impacts of their decisions on the problem. Methods employed in the "House Hunting Game", see later, are enabled by the development of such systems because multi user participation is produced whereby different people can take control of the IGS and thus impact upon the outcome or comment on other users of the system.

The IGS is an attempt to focus the research not on the methods of using GIS but what is actually being done with the technology. Thus non GIS information, which is relevant to the decision making process such as user background, political or private agendas be incorporated into an IGS system to enable a more rounded approach to spatial decision making. The modelling of the world in mathematical terms, such as co-ordinates and attributes is a valuable operation but when trying to "map" such information for decision making, it becomes un-stuck because it does not fully relate to the decision makers real world view. Other contextual information must be used to make sense of the data and information.

Through the use of an IGS, areas of consensus and conflict could be examined and the impacts of changes to decisions investigated and responses seen. This type of approach will require that there is a change in the culture of using GIS/computer systems whereby the user(s) perceives that the CSDM contains all of the measured information as well as their biases they have brought to the solution. However, the use of IGS systems may prove problematic because it challenges the status quo of existing system usage in that the CSDM is used as a creative as opposed to supportive tool.

An alternate approach, is that of the development of the meta spatial modelling system. A meta spatial model is a collection of the results of a series of other spatial models which can be integrated using for example, multicriteria analysis to determine a wider viewpoint of a problem domain(Heywood & Tomlinson, 1995). For example, an area could be modelled in terms of traffic flow, air pollution and noise pollution, these models could then be allocated quality measures and combined together with weightings determined by the user(s) using multicriteria analysis. Areas of conflict and consensus between the models could be produced and then used in decision making whether by changing the base models of the combination of the model results in the meta spatial model.

The meta spatial model makes use of a Geographical Operating Environment(GOE) which enables access to spatial and non spatial data and the functions associated with the data files. For example, a common work area can be defined in which the user can visualise a spreadsheet and a map and apply functions to both data files from the applications which created them without having to use the individual applications directly (Tomlinson 1994). Such an environment allows the limiting of the functions available to those that the user require. Function overload is very much part of GIS.

The meta spatial model is different from the IGS because it is specifically aimed at the development of systems which use modelling explicitly as the means of decision making. The IGS is a more flexible open ended environment which allows data/information exploration and is not specific to modelling as the means of decision making/analysis.

3.0 Real World Case Studies

The proposals above, taken from various papers, have been prototyped in two forms the "House Hunting Game" and the "Kirklees System", these are described below.

House Hunting Game - looking for a place to live

To demonstrate the applicability of multiparticipatory decision making using multicriteria analysis we have developed a simple to use computer system which is based upon looking for a place to live(Heywood, Oliver & Tomlinson, 1995). Multicriteria analysis has been selected because it is easy to understand in a non technical sense and forms part of the Idrisi functionality and is immediately available for use, any other algorithm could be used. When selecting a location to live, we are faced by decisions on the neighbourhood, insurance level, location of schools, modes of transport and many other issues. This system has been presented at conference workshops where people are grouped into families. Each family is assigned a character type, such as old retired, young couple and so on which they use to determine how they view the area and ultimately perceive the data.

The system developed, presents the family with a series of maps which represent the factors and constraints. The factors are numerous and include insurance and schools. For example the family may select to live in an area which has as low insurance price and is near a school. The constraints are for example "I don't want to live near a rubbish dump", "I don't want to live near a motorway".

Once they have decided how the factors and constraints fit with each other, they can use the system to produce a map showing the best and worst areas in which they would like to live by according weights to the factors and constraints. The interface to the system presents the family with maps representing the factors and constraints plus information on how the maps were formed and what they mean. The family then decide what is important for where they want to live and using a geographical equaliser they can alter the relative importance of factors and constraints which are then used in the analysis. The geographical equaliser is a tool using sliders bars to represent the various factors and their relative importance. The multicriteria analysis uses Idrisi and is performed on a real data set relating to an area of the County of Chesire, England.

The system enables group decision making to occur so that the family can see the effect of their decisions. They can then go back and alter their priorities, using the slider bars and see what the new map would look like. They can produce individual maps and compare the results to identify areas of potential conflict and consensus. This has been performed in rooms of 15 PC workstations where groups of people run the analysis and then examine each others results. It presents non GIS specialists with the opportunity to utilise GIS technology by non GIS methods - i.e. they family do not realise that they are using a GIS, they relate to the information. The use of the slider bar enables complex decision making to occur because the family can see the effect of the relative weights of the factors and constraints in a simple visual manner. This enables them to concentrate on the task of decision making rather than how to actually perform the decision making.

Work is current with Kirklees Metropolitan Council and local community groups in the village of Holmfirth, County of West Yorkshire, England to develop a consensus based decision making system. The Kirklees systems uses the same multicriteria analysis technique as the "House Hunting Game" for the assessment of local community attitudes towards their environment which can be used by the council for use in Local Agenda 21 proposed by the United Nations Agenda 21. The system is under development within initial trials in June 1995 and use in August 1995. The system will allow local people to identify what is important to them for their environment and weight these in order of importance to various development proposals from the Council. This will produce a series of personal maps which can be used to identify the conflict and consensus within the community with which the Council can then use to better develop policies for local sustainable development. The methods used are based upon group community usage of technology to understand their environment.

4.0 Conclusion

This paper has presented a brief view of our joint research work in the field of CSDM. The issues outlined below and those proposed through Initiative 17 are, we believe, important for the future development of CSDM.

Issues in CSDM that should be discussed at the workshop

  1. What areas of application are appropriate for CSDM research and how can we identify these ?

  2. How can we integrate various methods of data access into a single environment so that CSDM can take place ?

  3. What user interface mechanisms are appropriate ?

  4. What analytical techniques can be included within a CSDM environment ?

  5. What visual methods can be used to differentiate between ideas/data/information/decisions and should there be multiple methods of representation for multiple users preference ?

Through the production of prototypes and testing of ideas we can further develop the use of GIS techniques and technology to aid our understanding of the human and physical environments. The application of such techniques and technology should not remain in the domain of the specialist but be available to all who require such tools for decision making and understanding.

References

Densham P.J, (1991), Spatial Decision Support Systems: Chapter 26: Geographical Information Systems Eds Maguire Goodchild and Rhind), Longman, London.

Heywood.D.I & Carver.S.J,1994,Decision Support or Idea Generation: The role of GIS in policy formation, Angewandte Geographische Informationsverarbeitung VI, Salzburg

Heywood D.I, Oliver J & Tomlinson S.J, (1994), Building an exploratory multi criteria modelling environment for spatial decision support: Chapter 11, Innovations in GIS 2 Ed Fisher.P, Taylor & Francis, London

Heywood.D.I & Tomlinson S.J,(1995), Modelling Uncertainty, Consensus and Conflict: The Need for Spatial Meta Models in Environmental Decision Making, GISDATA Conference, Stockholm, Sweden.

Szajna.B & Scamell.R, (1993), The Effects of Information System User Expectations on Their Performance and Perceptions, MIS Quarterly, December 1993 pp 493-516

Wetherbe.J, (1991), Executive Information: Getting it Right, MIS Quarterly, March 1991 pp51-65

Joshi.K, (1991), A Model of Users' Perspective on Change: The Case of Information Systems Technology Implementation, MIS Quarterly, June 1991 pp 229-242

Lawerance.M & Low.G, (1993), Exploring Individual User Satisfaction within User-Led Development, MIS Quarterly, June 1993 pp195-209

Tomlinson S.J,(1994),Developing a Geographical Operating Environment (GOE),Decision Support 2001 Conference, Toronto.

Tomlinson S.J,(Thesis, in prep), Towards the development of a generic Geographical Operating Environment (GOE), Manchester Metropolitan University

Biographies

Dr Stephen J. Carver (steve@geography.leeds.ac.uk)
Dept of Geography
University of Leeds, Leeds, England

Dr Carver is currently a Lecturer in Geography in the School of Geography, University of Leeds,UK. Previous posts held include Research Associate at the University of Newcastle upon Tyne working within the NERC/ESRC Land Use Programme (NELUP) on developing planning and socio-economic aspects of the NELUP decision support system; University Research Fellow in Geographic Information Handling within the Centre for Urban and Regional Development Studies at the University of Newcastle upon Tyne working on methods of estimating error in digital map overlay operations and developing new error handling tools for GIS. His doctoral thesis focused on the application of GIS to siting radioactive waste disposal facilities in Britain with particular emphasis on developing workable links between GIS and multicriteria evaluation. Current research interests focus on environmental and social applications of GIS and on developing new methodologies for decision making and support.

Dr Ian Heywood (i.heywood@mmu.ac.uk)
Dept of Environmental & Geographical Sciences
The Manchester Metropolitan University, Manchester, England

Dr Heywood is currently a Senior Lecturer in the Department of Environmental & Geographical Sciences at The Manchester Metropolitan University, UK. Previous posts have included Lecturer at the University of Salford, Research Associate, University of Newcastle upon Tyne and Visiting Professor, Institute of Geography, University of Salzburg, Austria.. Dr Heywood is one of the co founders of the International Distance Learning Diploma/Msc in GIS, co-director of the Idrisi Resource Centre UK and member of the GISDATA initiative on Spatial Data Modelling. Research activities focus on Spatial Decision Support, Telematics and Distance Learning.

Steven J. Tomlinson (s.tomlinson@mmu.ac.uk)
Dept of Environmental & Geographical Sciences
The Manchester Metropolitan University, Manchester, England

Mr Tomlinson is a Lecturer in the Department of Environmental & Geographical Sciences at The Manchester Metropolitan University, UK. Previous posts have included Research Assistant at the University of Salford, UK working on a research project for Greater Manchester Fire Authority investigating the incidence of fire calls and the relevance of socio-economic factors used in decision making for the allocation of resources. His thesis ( in prep ) is an investigation into the production of a generic Geographical Operating Environment(GOE) which can be used for spatial an non spatial access for decision support. His research interests include Spatial Decision Support and the development of techniques/technology which make GIS more accessible to non technical users.

A combined list of some relevant references

Carver,S & Openshaw,S.(1995) Using GIS to explore the technical and social aspects of site selection. in Proceedings of Conference on the Disposal on the Geological Disposal of Radioactive Wastes, Royal Lancaster Hotel, London, March 1995

Carver.S, (1991) Integrating multicriteria evaluation with GIS, International Journal of Geographical Information Systems 5(3), 321-339

Carver.S, (1991) Spatial decision support systems for facility location: a combined GIS and multicriteria evaluation approach. in Proceedings of 2nd International Conference on Computers in Urban Planning and Urban Management, Oxford UK, July 1991, 75-90

Carver.S,(1991), A prototype decision support system for siting radioactive waste disposal facilities using Geographic Information Systems and multicriteria evaluation. North East Regional Research Laboratory Report No. 91/1. University of Newcastle upon Tyne.

Carver.S. Myers.A & Newton.M, (1992), The role of public perception in the response planning for major incidents: a proposal for a GIS-based strategy. North East Regional Research Laboratory Report No 92/1, University of Newcastle upon Tyne.