Developments in Spatial Data Handling

Spatial models are often based on polygons both in 2D and 3D. Many Geo-ICT products support spatial data types, such as the polygon, based on the OpenGIS ‘simple Features Specification’. OpenGIS and ISO have agreed to harmonize their specifications and standards. In this paper we discuss the relevant aspects related to polygons in these standards and compare several implementations. A quite exhaustive set of test polygons (with holes) has been developed. The test results reveal significant differences in the implementations, which causes interoperability problems. Part of these differences can be explained by different interpretations (definitions) of the OpenGIS and ISO standards (do not have an equal polygon definition). Another part of these differences is due to typical implementation issues, such as alternative methods for handling tolerances. Based on these experiences we propose an unambiguous definition for polygons, which makes polygons again the stable foundation it is supposed to be in spatial modelling and analysis. Valid polygons are well defined, but as they may still cause problems during data transfer, also the concept of (valid) clean polygons is defined.

The objective of GIS and Spatial Data Handling is to view, query and manipulate a computer simulation of the real world. While we traditionally work with two-dimensional static maps, modern technology allows us to work with a three-dimensional dynamic environment. We have developed a generic graphics component which provides many of the tools necessary for developing 3D dynamic geographical applications. Our example application is a 3D “Pilot Book”, which is used to provide navigation assistance to ships entering Hong Kong harbour. We show some of the “Marine GIS” results, and mention several other applications.

The Appalachian-Southern Africa Research and Development Collaboratory (ASARD) seeks to explore the integration of community decision-making with GIS across cultures. Combining geospatial data with local knowledge and the active participation of the community creates a Community-Integrated Geographic Information System (CIGIS) representing and valuing themes related to community and economic development. The intent is to integrate traditional GIS with the decision-making regime of local people and authorities to assist them in making informed choices and to increase local participation in land use planning, especially within economically disadvantaged communities.

Preference elicitation is a non-deterministic process that involves many intuitive and non well-defined criteria that are difficult to model. This paper introduces a novel approach that combines image schemata, affordance concepts and neural network for the elicitation of user’s preferences within a web urban space. The selection parts of the neural network algorithms are achieved by a web-based interface that exhibits image schemata of some places of interest. A neural network is encoded and decoded using a combination of semantic and spatial criteria. The semantic descriptions of the places of interest are defined by degrees of membership to predefined classes. The spatial component considers contextual distances between places and reference locations. Reference locations are possible locations from where the user can act in the city. The decoding part of the neural network algorithms ranks and evaluates reference locations according to user’s preferences. The approach is illustrated by a web-based interface applied to the city of Kyot

The general problem of retrieval and integration of data from a set of distributed heterogeneous data sources in response to a query has a number of facets. These include the breakdown of a query into appropriate subqueries that can be applied to different data sources as well as the integration of the partial results obtained to produce the overall result. The latter process is non-trivial and particularly dependent on the semantics of the data. This paper discusses an architecture developed to enable a user to query spatial data from a collection of distributed heterogeneous data sources. This has been implemented using GML to facilitate data integration. The system is currently being used to study the handling of positional uncertainty in spatial data in such a system.

Security is an important topic in the Information Systems and their applications, especially within the Internet environment. Security issue for geospatial data is a relatively unexplored topic in Geographical Information Systems (GIS). This paper analyzes the security solutions for

Geographical Information Storage Systems

(GISS) within n-tier GIS architecture. The first section outlines the application of the main categories of database security for management spatial data. These categories are then analyzed from a point of view of application within GIS. A

File System within Database

(FSDB) with traditional and new encryption algorithms has been proposed to be used as a new GISS solution. A FSDB provides more safe and secure storage for spatial files and support centralized authentication and access control mechanism in legacy DBMS. Cryptography solutions as a topic of central importance to many aspects of network security are discussed in detail. This part of the paper describes several traditional and new symmetric, fast and nonlinear encryption algorithms’ implementation with fixed and flexible key sizes.

The progressive transmission of map data over World Wide Web provides the users with a self-adaptive strategy to access remote data. It not only speeds up the web transfer but also offers an efficient navigation guide in information acquisition. The key technology in this transmission is the efficient multiple representation of spatial data and pre-organization on server site. This paper aims at offering a new model for the multiple representations of vector data, called

changes accumulation model

, which considers the spatial representation from one scale to another as an accumulation of the set of changes. The difference between two consecutive representations is recorded in a linear order and through gradually addition or subtraction of “change patches” the progressive transmission is realized. As an example, the progressive transmission of area features based on this model is investigated in the project. The model is built upon the hierarchical decomposition of polygon into series of convex hulls or bounding rectangles and the progressive transmission is accomplished through component of the decomposed elements.

Although the properties of natural neighbour interpolation and its usefulness with scattered and irregularly spaced data are well-known, its implementation is still a problem in practice, especially in three and higher dimensions. We present in this paper an algorithm to implement the method in two and three dimensions, but it can be generalized to higher dimensions. Our algorithm, which uses the concept of flipping in a triangulation, has the same time complexity as the insertion of a single point in a Voronoi diagram or a Delaunay triangulation.

An artificial and ‘real’ set of test data are modelled as continuous surfaces by linear interpolators and three different cubic interpolators. Values derived from these surfaces, of both elevation and slope, are compared with analytical values for the artificial surface and a set of independently surveyed values for the real surface. The differences between interpolators are shown with a variety of measures, including visual inspection, global statistics and spatial variation, and the utility of cubic interpolators for representing curved areas of surfaces demonstrated.

Here we present a contour-smoothing algorithm based on weighted smoothing splines for contour extraction from a triangular irregular network (TIN) structure based on sides. Weighted smoothing splines are one-variable functions designed for approximating oscillatory data. Here some properties are derived from a small space of functions and working with few knots and special boundary conditions. However, in order to apply these properties to a two variable application such as contour smoothing, local reference frames for direct and inverse transformation are required. The advantage of using weighted smoothing splines as compared to pure geometric constructions such as the approximation by parabolic arcs or other type of spline function is the fact that these functions adjust better to the data and avoid the usual oscillations of spline functions. We note that Bezier and B-spline techniques are result in convenient, alternative representations of the same spline curves. While these techniques could be adapted to the weighted smoothing spline context, there is no advantage as our approach will be simple enough.

We extend pit filling and basin hierarchy computation to TIN terrain models. These operations are relatively easy to implement in drainage computations based on networks (e.g., raster D8 or Voronoi dual) but robustness issues make them difficult to implement in an otherwise appealing model of water flow on a continuous surface such as a TIN. We suggest a consistent solution of the robustness issues, then augment the basin hierarchy graph with different functions for how basins fill and spill to simplify the watershed graph to the essentials. Our solutions can be tuned by choosing a small number of intuitive parameters to suit applications that require a data-dependent selection of basin hierarchies.

Topological relationships between spatial objects have been a focus of research on spatial data handling and reasoning for a long time. Especially as predicates they support the design of suitable query languages for data retrieval and analysis in spatial databases and geographical information systems. Whereas research on this topic has always been dominated by qualitative methods and by an emphasis of a strict separation of topological and metric, that is, quantitative, properties, this paper investigates their possible coexistence and cooperation. Metric details can be exploited to refine topological relationships and to make important semantic distinctions that enhance the expressiveness of spatial query languages. The metric refinements introduced in this paper have the feature of being topologically invariant under a ne transformations. Since the combination of a topological predicate with a metric refinement leads to a single unified quantitative measure, this measure has to be interpreted and mapped to a lexical item. This leads to vague topological predicates, and we demonstrate how these predicates can be integrated into a spatial query language.

In this article we apply fuzzy bspline reconstruction supplemented by fuzzy kriging to the problem of constructing a smooth deterministic model for environmental pollution data. A method to interrogate the model will also be discussed and applied.

Handling uncertain spatial data and modelling of spatial data quality and data uncertainty are currently major challenges in GIS. Geodata usage is growing, for example in agricultural and environmental models. If the data are of a low quality, then model results will be poor as well. An important issue to address is the accuracy of GIS applications for model output. Spatial data uncertainty models, therefore, are necessary to quantify the reliability of model results. In this study we use a combination of fuzzy methods within geostatistical modelling for this purpose. The main motivation is to jointly handle uncertain spatial and model information. Fuzzy set theory is used to model imprecise variogram parameters. Kriging predictions and kriging variances are calculated as fuzzy numbers, characterized by their membership functions. Interval width of predictions measures the effect of variogram uncertainty. The methodology is applied on methane (CH

4

) emissions at the Island of Java. Kriging standard deviations ranged from 12 to 26.45, as compared to ordinary kriging standard deviations, ranging from 12 to 33.11. Hence fuzzy kriging is considered as an interesting method for modeling and displaying the quality of spatial attributes when using deterministic models in a GIS.

At the end of the sixties Hägerstrand introduced a space-time model which included features such as a Space-Time-Path, and a Space-Time-Prism. From a visualization perspective the Space-Time-Cube was the most prominent element in Hagerstrand’s a pproach. However, when the concept was introduced the options to create the graphics were limited to manual methods and the user could only experience the single view created by the draftsperson. Today’s software has options to automatically create the cube and its contents from database. Data acquisition of space-time paths for both individuals and groups is also made easier using GPS. The user’s viewing environment is, by default, interactive and allows one to view the cube from any direction. In this paper the visualization environment is proposed in a geovisualization cont

Technological advances in position aware devices increase the availability of tracking data of everyday objects such as animals, vehicles, people or football players. We propose a geographic data mining approach to detect generic aggregation patterns such as flocking behaviour and convergence in geospatial lifeline data. Our approach considers the object’s motion properties in an analytical space as well as spatial constraints of the object’s lifelines in geographic space. We discuss the geometric properties of the formalised patterns with respect to their efficient computation.

In a context-aware environment, the system must be able to refresh the answers to all pending queries in reaction to perpetual changes in the user’s context. This added to the fact that mobile systems suffer from problems like scarce bandwidth, low quality communication and frequent disconnections, leads to high delays before giving up to date answers to the user. A solution to reduce latency is to use hoarding techniques. We propose a hoarding policy particularly adapted for location-dependent information systems managing a huge amount of multimedia information and where no assumptions can be made about the future user’s location. We use the user’s position as a criterion for both hoarding and cache invalidation.

Geographic Information Retrieval is concerned with retrieving documents that are related to some location. This paper addresses the ranking of documents by both textual and spatial relevance. To this end, we introduce

distributed ranking

, where similar documents are ranked spread in the list instead of consecutively. The effect of this is that documents close together in the ranked list have less redundant information. We present various ranking methods, efficient algorithms to implement them, and experiments to show the outcome of the methods.

In earlier work, we introduced the notion of the F-histogram and demonstrated that it can be of great use in understanding the spatial organization of regions in images. Moreover, we have recently designed F-histograms coupled with mutually exclusive and collectively exhaustive relations between line segments. These histograms constitute a valuable tool for extracting topological relationship information from 2D concave objects. For any direction in the plane, they define a fuzzy partition of all object pairs, and each class of the partition corresponds to one of the above relations. The present paper continues this line of research. It lays the foundation for generating a linguistic description that captures the essence of the topological relationships between two regions in terms of the thirteen Allen relations. An index to measure the complexity of the relationships in an arbitrary direction is developed, and experiments are performed on real data.

This document describes the Po-tree, a new indexing structure for spatiotemporal databases with real-time constraints. Natural risks management and other system can use arrays of spatially referenced sensors, each of them sending their measurements to a central database. Our solution tries to facilitate the indexing of these data, while favoring the newer ones. It does so by combining two sub-structures for the spatial and temporal components. While Mobility is not yet supported, evolutions of the structures shall be able to deal with it.

Humans perceive the boundary of locality vaguely. This paper presents how the indeterminate boundaries of localities can be represented in GIS. For this task, indeterminate boundaries of localities are modeled by a fuzzy set membership function in which generic rules on geospatial objects are incorporated. Georeferenced traffic crash data reveal that police officers identify localities precisely at best 88% of the time. An empirical analysis indicates that people are 6% more confident in identifying urban localities than rural localities. As a conclusion, fuzzy set theory seems to provide a reasonable mechanics to represent vague concept of geospatial objects. The comparison of urban versus rural localities with respect to location indeterminacy suggests that neighborhood types may affect the way humans acquire spatial knowledge and forge mental representations of it.

Reference control points (RCPs) used to establish the regression model in registration or geometric correction are commonly assumed “perfect”. However, this assumption is often violated in practice due to RCPs actually always containing errors. Moreover, the errors in RCPs are one of main sources lowering the accuracy of geometric correction of uncorrected image. In this case Ordinary least squares (OLS) estimator, widely used in geometric correction of remotely sensed data, is biased and does not have the ability to handle explanatory variables with error and to propagate appropriately errors from RCPs to the corrected image. In this paper, we introduce the consistent adjusted least squares (CALS) estimator and propose a relaxed consistent adjusted least squares (RCALS) method, which can be applied to more general relationship, for geometric correction or registration. These estimators have good capability in correcting errors contained in the RCPs, and to propagate correctly errors of the RCPs to the corrected image with and without prior information. The objective of the CALS and our proposed RCALS estimators is to improve the accuracy of measurement value by weakening the measurement errors. For validating CALS and RCALS estimators, we employ the CALS and RCALS estimators using real-life remotely sensed data. It has been argued and demon-strated that CALS and RCALS estimators give superior overall performances in estimating the regression coefficients and variance of measurement error.

In the identification of landscape features vagueness arises from the fact that the attributes and parameters that make up a landscape vary over space and scale. In most existing studies, these two kinds of vagueness are studied separately. This paper investigates their combination (double vagueness) in the identification of coastal landscape units. Fuzzy set theory is used to describe the vagueness of geomorphic features based on continuity in space. The vagueness resulting from the scale of measurement is evaluated by statistical indicators. The differences of fuzzy objects derived from data at differing resolutions are studied in order to examine these higher-order uncertainties. Multi-scale analysis of the landscape is carried out using a moving window, ranging in size from 60x60 meters to 1500x1500 meters. The statistics of the fuzziness of the fuzzy landscape units are calculated, and the variability of them with scale is assessed. It shows that a major affect of scale on the mapping of geomorphic landscape units, is determination of the area of those units. This result implies that caution must be exercised in comparing landscapes at different scales and in choosing the resolution of the data that best describes the process under study.

This paper focuses on crisp and fuzzy operators to compute the area and perimeter of fuzzy geographical entities. The limitations of the crisp area and perimeter operators developed by Rosenfeld (1984) are discussed, as well as the advantages of the fuzzy area operator developed by Fonte and Lodwick (2004). A new fuzzy perimeter operator generating a fuzzy number is proposed. The advantage of using operators generating fuzzy numbers is then illustrated by computing the shape of a FGE, through its compactness, using the extension principle and the fuzzy area and perimeter.

This paper presents our proposal to evaluate generalised data from non generalised data. We propose a methodology of evaluation that decomposes the process of evaluation into sub-processes where the user can enter his own specifications and criteria of evaluation. We also propose different levels of synthesis according to the evaluation needs. The evaluation can be an ‘evaluation for editing’ to detect and correct generalisation inconsistencies, a ‘descriptive evaluation’ to describe more precisely how the data set represents the reality, or a ‘evaluation for marking’ to compare different generalisations. Of course working on data evaluation is also very useful to improve generalisation software. The proposed methodology is implemented on Laser-Scan Lamps2 GIS and has been tested on real geo-graphical data base (the IGN-France BDTopo©) to produce maps at medium scale.

Road network generalization aims to simplify the representation of road networks by reducing details, while maintaining network connectivity and overall characteristics. This paper presents a method to select salient roads based on connection analysis. The number of connections is counted at each junction, which acts as a parameter indicating the association between salient roads. Several strategies are proposed to combine connection criterion with road length and road attributes (when available), so as to order roads corresponding to their relative importance in road network generalization. A case study shows that connection criterion is favored for maintaining the density differences in road network, and the combined criteria creates more reasonable results than only length criterion.

Visualizing spatial information on small mobile displays is a big chance and challenge at the same time. In order to address the tradeoff between huge spatial data sets and small storage capacities and visualization screens, we propose to visualize only the information on the screen which adequately fits the current resolution. To realize this, we automatically decompose the generalization of an object into a sequence of elementary steps. From this, one can later easily obtain any desired generalization level by applying the appropriate subpart of the sequence. The method method does not only lead to smooth transitions between different object representations but also is useful for incremental transmission of maps through limited bandwidth channels.

Few line generalisation algorithms provide explicit control over the style of generalisation that results. In this paper we introduce weighted effective area, a set of area-based metrics for cartographic line generalisation following the bottom-up approach of the Visvalingam-Whyatt algorithm. Various weight factors are used to reflect the flatness, skewness and convexity of the triangle upon which the Visvalingam-Whyatt effective area is computed. Our experimental results indicate these weight factors may provide much greater control over generalisation effects than is possible with the original algorithm. An online web demonstrator for weighted effective area has been set up.

This paper introduces a reasoning system based on ternary projective relations between spatial objects. The model applies to spatial objects of the kind point and region, is based on basic projective invariants and takes into account the size and shape of the three objects that are involved in a relation. The reasoning system uses permutation and composition properties, which allow the inference of unknown relations from given ones.

Even though the positions of objects may be uncertain, one may know some topological information about them. In this paper, we develop a new model for storing topological relationships in uncertain spatial data. It is intended to be the equivalent of such representations as the Node-Arc-Area representation but for spatial objects with uncertain positions.

Two topological properties of raster region — connectedness and perforation — are examined in the context of spatial optimization. While topological properties of

existing

regions in raster space are well understood,

creating

a region of desired topological properties in raster space is still considered as a complex combinatorial problem. This paper attempts to formulate constraints that guarantee to select a connected raster region with specified number of holes in terms amenable to mixed integer programming models. The major contribution of this paper is to introduce a new intersection of two areas of spatial modeling — discrete topology and spatial optimization — that are generally separate.

The

theory theory

claims that children’s acquisition of knowledge is based on forming and revising theories, similar to what scientists do (Gopnik and Meltzoff 2002). Recent findings in developmental psychology provide evidence for this hypothesis.

Children have concepts about space that differ from those of adults. During development these concepts undergo revisions.

This paper proposes the formalization of children’s theories of space in order to reach a better understanding on how to structure spatial knowledge. Formal models can help to make the structure of spatial knowledge more comprehensible and may give insights in how to build GIS. Selected examples for object appearances are modeled using an algebra. An

Algebra Based Agent

is presented and coded in a functional programming language as a simple computational model.

This paper presents a methodology to generate a topologically correct street centreline from city block boundaries using ArcGIS software. The approach utilises Sugihara’s (1992) point approximation algorithm as a starting point to create an area Voronoi diagram which forms the basis of the centreline. A recursive method is introduced to schematize the geometry of the Voronoi medial axis. The approach is applied to data from the City of Rosario, Argentina. The paper concludes with suggestions for further enhancements to the approach that have, among other things, the potential to automate attribution to network segments using adjacent polygon attributes.

Traffic data is the most important component of any transport GIS. They are mainly collected by traffic sensors (detectors). However, the installation of sensors is quite expensive. There have been studies on the finding of optimum location of inductive loop detector, the most widely used sensor in the past decades. However, with the advancement of new sensor technologies in recent years, many new sensors are now available for multi-lane and multi-direction traffic monitoring. Most of them are nonintrusive sensors which are more suitable be located at road junctions instead of roadways. Thus, different from previous studies on link-based sensor location, this paper explores method in determining critical road network junctions for the optimum location of nonintrusive sensors for monitoring and collecting real-time traffic data. The objective is to select the least number of junctions while maximally cover the road network. Since the problem is NP-complete, a greedy-based heuristic method is proposed and a numerical experiment is conducted to illustrate its efficiency.

Although geographical information system (GIS) is being used as a tool in Malaysian development process, this is being limited to the planning and design stage of the process. There have been very few Malaysian initiatives to further the usage of GIS into the operation and post-operation phases of the development process. This paper looks at the effort by International Islamic University Malaysia (IIUM) in developing GIS-based system to assist the management of her campus.

Landmark-based navigation is the most natural concept for humans to navigate themselves through their environment. It is desirable to incorporate this concept into car and personal navigation systems, which are nowadays based on distance and turn instructions. In this paper, an approach to identify landmarks automatically using existing GIS databases is introduced. By means of data mining methods, building information of the digital cadastral map of Germany is analyzed in order to identify landmarks. In a second step, a digital surface model obtained by laser scanning is used to assess the visibility of landmarks for a given route.

Route planners are tools that support the navigator in selecting the best route between two locations. Solving a route choice problem involves sorting and ranking of alternatives according to underlying evaluation criteria and decision rules. Using an appropriate classification of route selection criteria in the user interface is an important ingredient for user friendly route planners. The paper presents a method for assessing a hierarchical structure of route selection criteria for bicycle route planning tasks along with data from two empirical studies. The first study investigates route selection criteria that are relevant for bicycle navigation in urban environments. The second study reveals preferred classification schemata for these criteria. The presented methodology can be adopted for other transportation domains, such as car or pedestrian navigation.

An algorithm has been developed for icon labelling on a real-time map. The algorithm is based on a least-disturbing definition and positions the icons in an area where they obscure the cartographic data as little as possible. To find this area the algorithm performs a spiral search on a precomputed grid. The computational complexity is low, which makes it possible to use the algorithm in real-time applications. A case study has been performed which demonstrates the improvement in icon labelling achieved by the algorithm.

This paper presents an approach for a semantically correct integration of a 2.5D digital terrain model (DTM) and a 2D topographic GIS data set. The algorithm is based on a constrained Delaunay triangulation. The polygons of the topographic objects are first integrated without considering the semantics of the object. Then, those objects which contain implicit height information are dealt with. Object representations are formulated, the object semantics are considered within an optimization process using equality and inequality constraints. First results are presented using simulated and real data.

A lot of attention has been paid to generalization (filtering) of Digital Elevation Models (DEMs) and the same is true for generalization of 2D object models (e.g. topographic or land use data). In addition there is a tendency to integrate DEMs with classified real-world objects or features, the result is sometimes called a Digital Terrain Model (DTM). However, there has not been much research on the generalization of these integrated elevation and object models. This paper describes a four step procedure. The first two steps have been implemented and tested with real world data (laser elevation point clouds and cadastral parcels). These tests have yielded promising results as will be shown in this paper.

A model of automatically detecting the building shadows in high resolution aerial remote sensing image is introduced in this paper. The space coordinates of the shadows are first computed using photogrammetric engineering. To do this, digital surface model (DSM) and the sun zenith and azimuth angles are used. By camera model, the scanning line and the camera position are calculated for each space shadow. A contour driven height field ray tracing is proposed to determine the visibility of a shadow. For a visible shadow, its projecting in the image, called measured shadow, is calculated by collinearity equations. Then by image analysis the reference segmentation threshold is obtained from the intensity distribution of the measured shadows. At last, the image segmentation is implemented to get the precise image shadow areas.

This paper presents a formalized ontological framework for the analysis of classifications of geographic objects. We present a set of logical principles that guide geographic classifications and then demonstrate their application on a practical example of the classification of ecosystems of Southeast Alaska. The framework has a potential to be used to facilitate interoperability between geographic classifications.

Expressions of expert opinion are being used to relate ontologically diverse data and to identify logical inconsistency them. Relations constructed under different scenarios, from different experts and evidence combined in different ways identify different subsets of inconsistency, the reliability of which can be parameterised by field validation. It is difficult to identify one combination as being objectively “better” than other. The selection of specific experts and scenarios depends on user perspectives.

This paper describes a programmatic framework for representing, manipulating and reasoning with geographic semantics. The framework enables automating tool selection for user defined geographic problem solving, and evaluating semantic change in knowledge discovery environments. Methods, data, and human experts (our resources) uses, inputs, outputs, and semantic changes are described using ontologies. These ontological descriptions are manipulated by an expert system to select resources to solve a user-defined problem. A semantic description of the problem is compared to the services that each entity can provide to construct a graph of potential solutions. An optimal (least cost) solution is extracted from these solutions, and displayed in real-time. The semantic change(s) resulting from the interaction of resources within the optimal solution are determined via expressions of transformation semantics represented within the Java Expert System Shell. This description represents the formation history of each new information product (e.g. a map or overlay) and can be stored, indexed and searched as required. Examples are presented to show (1) the construction and visualization of information products, (2) the reasoning capabilities of the system to find alternative ways to produce information products from a set of data methods and expertise, given certain constraints and (3) the representation of the ensuing semantic changes by which an information product is synthesized.

The notion of data quality is of particular importance to geographic data. One reason is that such data is often inherently imprecise. Another is that the usability of the data is in large part determined by how “good” the data is, as different applications of geographic data require different qualities of the data are met. Such qualities concern the object level as well as the attribute level of the data. This paper presents a systematic and integrated approach to the conceptual modeling of geographic data and quality. The approach integrates quality information with the basic model constructs. This results in a model that enables object-oriented specification of quality requirements and of acceptable quality levels. More specifically, it extends the Unified Modeling Language with new modeling constructs based on standard classes, attributes, and associations that include quality information. A case study illustrates the utility of the quality-enabled model.

There currently exist many geographical databases that represent a same part of the world, each with its own levels of detail and points of view. The use and management of these databases therefore sometimes requires their integration into a single database. The main issue in this integration process is the ability to analyse and understand the differences among the multiple representations. These differences can of course be explained by the various specifications but can also be due to updates or errors during data capture. In this paper, we propose an new approach to interpret the differences in representation in a semiautomatic way. We consider the specifications of each database as the “knowledge” to evaluate the conformity of each representation. This information is grasped from existing documents but also from data, by means of machine learning tools. The management of this knowledge is enabled by a rule-based system. Application of this approach is illustrated with a case study from two IGN databases. It concerns the differences between the representations of traffic circles.

This paper extends upon a category of information, processes description, and its relevance in metadata about geographic information. Metadata bases about processes are needed because processes are themselves resources to manage. Besides, specific processes participate in the description of other types of resources like data sets. Still, current metadata models lack structured containers for this type of information. We propose a model to build structured descriptions of processes to be integrated in metadata bases about processes themselves or about data sets.

We are concerned with comparing two or more categorical maps. This type of task frequently occurs in remote sensing, in geographical information analysis and in landscape ecology, but it is also an emerging topic in medical image analysis. Existing approaches are mostly pattern-based and focus on composition, with little or no consideration of configuration. Based on a web-survey and a workshop, we identified some key strategies to handle local and hierarchical comparisons and developed algorithms which include significance tests. We attempt to fully integrate map comparison in a process-based inferential framework, where the critical questions are: (1) Could the observed differences have arisen purely by chance? and/or (2) Could the observed maps have been generated by the same process?

One of the major research areas in geovisualization is the exploration of patterns and relationships in large datasets for understanding underlying geographical processes. One of the attempts has been to use Artificial Neural Networks as a technology especially useful in situations where the numbers are vast and the relationships are often unclear or even hidden. We investigate ways to integrate computational analysis based on the Self-Organizing Map, with visual representations of derived structures and patterns in an exploratory geovisualization environment intended to support visual data mining and knowledge discovery. Here we explore a large dataset on health statistics in Africa.

Disease rates computed for small areas such as zip codes, census tracts or census block groups are known to be unstable because of the small populations at risk. All people in Iowa diagnosed with colorectal cancer between 1993 and 1997 were classified by cancer stage at the time of their first diagnosis. The ratios of the number of late-stage cancers to cancers at all stages were computed for spatial aggregations of circles centered on individual grid points of a regular grid. Late-stage colorectal cancer incidence rates were computed at each grid point by varying the size of the spatial filter until it met a minimum threshold on the total number of colorectal cancer incidences. These different-sized areas are known as spatially adaptive filters. The variances analyzed at grid points showed that the maps produced using spatially adaptive filters gave higher statistical stability in computed rates and greater geographic detail when compared to maps produced using conventional fixed-size filters.