Mark Gahegan

Introducing GeoVISTA Studio: an integrated suite of visualization and computational methods for exploration and knowledge construction in geography

One barrier to the uptake of geocomputation is that, unlike GIS, it has no system or toolbox that provides easy access to useful functionality. This paper describes an experimental environment, GeoVISTA Studio, that attempts to address this shortcoming. Studio is a Java-based, visual programming environment that allows for the rapid development of complex data exploration and knowledge construction applications to support geographic analysis. It achieves this by leveraging advances in geocomputation, software engineering, visualization and machine learning. At the time of writing, Studio contains full 3D rendering capability and has the following functionality: interactive parallel coordinate plots, scatterplot, visual classifier, 2D map and image viewer, sophisticated colour selection (including Munsell colour-space), spreadsheet, statistics package, and supervised and unsupervised neural networks. Through examples of Studio at work, this paper demonstrates the roles that geocomputation and visualization can play throughout the scientific cycle of knowledge creation, emphasising their supportive and mutually beneficial relationship. A brief overview of different types of inference used in such knowledge creation activities is given, and related to the exploratory analysis tools described. By way of results, a detailed account of the use of these tools is presented, and various findings and insights generated are pointed out. The domain of application is the process of uncovering useful categories by which a taxonomy of landuse/landcover can be created. The proposed categories are then evaluated using a combination of neural and visual methods, to ensure their viability.

Proximity operators for qualitative spatial reasoning

One way to increase the power of Qualitative Spatial Reasoning is to introduce proximity operators (such as close and far) that are surrogates for distance measures. These operators appear to be semi-quantitative in nature as opposed to purely qualitative. In the light of observations drawn from psychometric testing of perceived proximity, this paper discusses how a model to support proximal reasoning could be constructed. The relationships between the model and the raw data are described. Fuzzy set membership is used to reason about the degree of closeness. The formulation of queries involving proximity is presented, with the meaning of linguistic variables being instantiated within a given context at execution time.

CONSUMPTION OF DIFFERENT ALCOHOLIC BEVERAGES AS PREDICTORS OF LOCAL RATES OF NIGHT-TIME ASSAULT AND ACUTE ALCOHOL-RELATED MORBIDITY

Objective: To determine whether population levels of consumption of some alcoholic beverages are more closely associated with levels of harm than others, particularly if consumption of cask wine is more strongly related to rates of acute alcohol problems than consumption of bottled wine as a consequence of the extremely low rates of federal tax levied on the former. Method: A database of alcohol consumption and related problems was established for 130 areas of Western Australia. Demographic and economic data for these areas were included from the 1991 census. Empirically derived assumptions regarding the mean wholesale price of cask and bottled wine were utilised. Regression analyses examined the extent to which the consumption of different alcoholic beverages predicted levels of major varieties of harm. Results: Only cask wine and high‐strength beer consumption were significantly associated with rates of night‐time assault; consumption of all beverage varieties except bottled wine was significantly associated with rates of acute alcohol‐related morbidity. Further analyses, which included controls for an effect of total alcohol consumption, confirmed the pronounced contributions of cask wine and high‐strength beer to rates of night assaults and acute alcohol‐related morbidity. The proportion of all alcohol consumed as low‐alcohol beer was significantly negatively associated with these harms. Conclusions: The beverages most associated with rates of night‐time assaults and acute alcohol‐related morbidity are those with the lowest federal taxation per standard drink, i.e. cask not bottled wine and regular‐strength not low‐alcohol beer.

Alcohol consumption and injury in Western Australia: a spatial correlation analysis using geographic information systems

Abstract: Geographic information systems technology was used to describe, in geographical terms, the nature and strength of the relationship in Western Australia between alcohol consumption and the rates of related injury: night‐time assaults (10 p.m. to 6 a.m.); minor night‐time road crashes (10 p.m. to 6 a.m.), weighted by traffic density; and hospital E‐code (external‐cause) morbidity, weighted by alcohol aetiologic fractions. The data were aggregated by five conventional state regions: northern, central, western, southern and Perth metropolitan. There was a general association, of equal significance for males and females, between estimated per capita alcohol consumption and the selected rates of injury in the five regions. However, the nature and strength of association between alcohol consumption and individual injury measures varied. Night‐time assaults and hospital E‐code morbidity were strongly associated with consumption. Minor night‐time crashes had only a weak association. The variation in the relationship between alcohol consumption and injury suggests that prevention strategies need to take into account the particular drinking patterns and associated harm that occur in different regions of the state, and to develop a range of targeted responses. High rates of consumption and injury in most country areas support the need for greater regional prevention efforts.

Scatterplots and scenes : Visualisation techniques for exploratory spatial analysis

A collection of geographic data from a particular region contains many explicit and implicit relationships which are difficult to display and communicate without resorting to statistical summarisation or fragmentation into themes. This paper addresses the problems of exploring interconnections between layers of data using state-of-the-art visualisation techniques and is based on the premise that visual exploratory data analysis is a useful tool for providing insight into the complex and subtle relationships that occur in geography. The tools and techniques described extend beyond the current capacity of commercial geographic information systems in terms of (a) the flexibility of the scene description, (b) the volume of data (particularly the number of distinct layers or themes that are viewable concurrently) and (c) the facilities by which the user may study relationships. The techniques proposed are justified from a psychometric standpoint and some important limitations with their use are described. As well as the pictures shown in the paper, further results are given in the form of high resolution colour images, VRML scenes and video clips which may be downloaded from an accompanying Web site.

Distinguishing Instances and Evidence of Geographical Concepts for Geospatial Database Design

In many geoscientific disciplines concepts are regularly discovered and modified, but the architecture of our geospatial information systems is primarily aimed at supporting static conceptual structures. This results in a semantic gap between our evolving understanding of these concepts and how they are represented in our systems. The research reported here provides better database support for geographical concepts that evolve with particular situations. To reduce the potential for schema change in such environments, we develop an analysis of the structure and function of situated geographical concepts and directly model the results in an UML schema. The developed schema explicitly contextualizes geographic information and concepts, enabling the extraction of contexts and interpretations from databases. This aids (1) the uncovering of the implicit aspects of data, (2) the addition of empirical components to geoscientific ontology, and (3) enhances the context represented in geo-databases. Prototype implementations that show promise for managing geoscientific ontologies and databases are also briefly discussed.

Characterizing the Semantic Content of Geographic Data, Models, and Systems

Geographic information systems (GIS) employ distinct conceptual models of geographic space (Goodchild 1992), often as a reflection of the origins of the software (e.g., computer-assisted design and image processing). Some of these models are radically different, such as the images employed by Idrisi compared to the object coverages used by ARC/INFO. Others are more subtly different, such as a topologically-oriented coverage compared to the spaghetti polygons used by many desktop GIS. The meaning of spatial data is not the same within these models (Morehouse 1990), and translation or interoperation that is based solely on the geometry can lead to logical inconsistencies within the translated data. Whilst a good deal of very useful progress has been made by the likes of ISO TC211 and the Open Geodata Interoperability Specification (and related models), as yet these standards fall somewhat short in addressing the semantics of the underlying geographic models. Efforts thus far have tended to concentrate on geometry, which can be rather ambiguous in defining the meaning of data in a geographic sense.

Specifying the transformations within and between geographic data models

Geographic information is acquired according to several different underlying models of geographic space. Any meaning inherent within a dataset is intrinsically connected to the model by which it was captured. A design is presented for an integrated geographic information system offering a number of distinct views onto geographic space, of varying degrees of abstraction, to support the needs of an expanded user base. An architecture for such a system is developed by an extension of the traditional three layer architecture used in database design. The movement of data from one level of abstraction to another is formalized by a series of dataset and model transformations operating between four different geographic data models. From this formalism, a functional taxonomy of GIS operations is developed.

The integration of scene understanding within a Geographic Information System: A prototype approach for agricultural applications

This paper describes progress towards engineering a combined GIS and image understanding system, designed to analyse remotely-sensed imagery in a task-oriented manner, with the task and contextual parameters being supplied by the GIS. The overall aim is to evaluate some of the more recent developments in the field of computer vision and to show how they may be integrated (along with more traditional methods of image interpretation) to improve system performance in terms of object recognition and extraction. Three aspects are reported: (1) Development of a formal notation as the basis of describing the process of transforming data from low level image representations to high level object representations; (2) Design of collaborating low and high level image processes used in this abstraction process and their control via Process Networks encapsulated into frames; and (3) Use of Dempster-Shafer Decision Theory as the basis for combining different types of evidence to improve the recognition of objects in the scene. Some sample results are presented to illustrate the issues raised and mechanisms for the co-operative processing of data are suggested.

Experiments Using Context and Significance to Enhance the Reporting Capabilities of GIS

The use of a contextual reporting technique, to supplement the basic display of geographic data has been previously described (Gahegan, 1994; 1995). Standard forms of graphical and textual reporting are embellished with deductions regarding the significance of reported values, using automated reasoning to examine data within a defined context. This paper explores some of the different types of reporting behaviour that can be made available within a GIS, by defining context in terms of the geographic properties of data. Such a context may be defined over any group of geographic objects, and then specialised, using constraints in the spatial and temporal domains, or generalised, using the class specialisation hierarchy to examine progressively broader contexts. A brief formalism for geographic context is developed. Results are presented using a dataset of agricultural land and these show the adaptive nature of the reporting behaviour and demonstrate some of its potential uses. A discussion of the strengths and weaknesses of the approach is also given.