Annotated Bibliography : Visualization
Alan M. MacEachren and Menno-Jan Kraak
Exploratory Cartographic Visualization: Advancing the Agenda
Computers and Geosciences Vol. 23, No. 4, 335-343, 1997
Visualization of georeferenced data is presented. The presented approach emphasizes the use of visual methods in decision making. The ‘links’ between ‘cartographic visualization’ and scientific visualization are discussed.
The framework for visualization is described through the Cartography-3 model which characterize map space as a 3-D space. In this space, visualization is considered to be the complement of communication. The axes of the space are: human-interaction (high to low), knowledge extraction/persuasion and audience.
The goals of map use are also presented in a 3-D graphical model with respect to the 3-axes of Cartography-3 model. The goals are to: present, synthesize, analyze and explore. The synthesizing and exploration are more important to derive meaning from the data.
Web: http://www.gis.psu.edu/ica/ICAvis.html
Alan M. MacEachren
VISUALIZATION – Cartogrpahy for the 21st century.
International Cartographic Association Commission of Visualization’s website
http://www.geog.psu.edu/ica/icavis/poland1.html
Go to Top
This paper provides a brief overview of the evolution of Geographic Visualization (GVis) and the activities of the ICA Commission on Visualization. Initial research objectives for the Commission are detailed and a draft research agenda being developed is outlined.
The 1987 US NSF report characterized visualization as a method (and a product) that integrates the power of digital computers and human vision and directs the result toward facilitating scientific insight.
In 1995, the ICA approved formation of a new Commission on Visualization. The commission has developed Cartography-3 Model (elaborated in MacEachren and Kraak 1997).
Alan M. MarEachren, Monica Wachowicz, Robert Edsall and Daniel Haug
Constructing knowledge from multivariate spatiotemporal data: integrating geographical visualization with knowledge discovery in database methods.
International Journal of Geographic Information Science, 1999, Vol. 13, No. 4, 311-334
First, both Geographic Visualization (GVis) and Knowledge Discovery in Database (KDD) are defined. Methods from these two source domains are integrated. Next, recent GVis and KDD developments are reviewed and compared and potential for their integration is considered, emphasizing that an iterative process with user interaction is a central focus for uncovering interest and meaningful patterns through each. Then an approach to design of an integrated GVis and KDD environment directed to exploration and discovery is introduced. The approach emphasizes a matching of GVis and KDD meta-operations.
There is a good literature review on geographic visualization.
Web: http://www.geovista.psu.edu/ijgis.htm
Antony Unwin
Exploring spatio-temporal data
Chapter seven, Data Mining, ed.
Although the paper discusses exploratory data analysis, there are some elements of visualization. Analysis of time component is especially highlighted. Tools employed are interactive – interrogation, editing, overlaying, aligning, zooming, rescaling, smoothing and transformation. The software system REGARD is used for time data series.
Daniel A. Keim and Hans-Peter Kriegel
Using Visualization to Support Data Mining of Large Existing Databases
Ref: ????
This paper presents ideas how visualization technology can be used to improve the difficult process of querying very large databases. For this purpose a software system ‘VisDB’ is presented. The VisDB system tries to provide visual support not only for the query specification process, but also for evaluating query results and, thereafter, refining the query accordingly. By arranging and coloring the pixels according to the revelvance for the query, the user gets a visual impression of the resulting data set and of its relevance for the query. Using an interactive query interface, the user amy change the query dynamically and receives immediate feedback by the visual representation of the resulting data set. By using multiple windows for different parts of the query, the user gets visual feedback for each part of the query and, therefore, may easier understand the overall result.
This query interface is generally useful for specifying SQL-like queries. Data mining is viewed as an interactive hypothesis-generation process – "goal is to get the data to ask questions, rather than asking questions to the data."
David J. Hand
Data Mining: Statistics and More?
The American Statistician, May 1998, Vol. 52, No. 2
This paper is written from the viewpoint of statistics. The author points to some pitfalls, which should be taken care of. The aim of data mining is to identify patterns. The trouble is that many of these "patterns" will simply be a product of random fluctuations, and will not represent any underlying structure. The object of data analysis is not to model the fleeting random patterns of the moment, but to model the underlying structures which give rise to consistent and replicable patterns.
The main difference in methods of data mining and statistics lies in the sheer size of the data sets now available. In general, statistical methods are applied to primary data with predefined questions in mind whereas in data mining secondary data analysis is applied with no prior knowledge of what is being sought.
In large data sets, pattern searches will throw up a large number of candidate patterns, there will be a high probability that spurious (chance) data configurations will be identified as patterns. It is only possible that a solution will only be found by stepping outside the conventional probabilistic statistical framework.
Demin Xiong (*****)
Traffic Flow Modeling with Visualization Tools
GIS-LIS ’92 Annual Conference Proceedings
This research is conducted from a GIS perspective that places emphasis upon the development of methods to integrate modeling and visualization techniques in order to analyze traffic flows in more effective ways. This paper first introduces a conceptual framework which can be used to integrate various procedures for effective and efficient traffic flow modeling and visualization. Then techniques to carry on model calculation with user-equilibrium modeling as a specific example are presented. Finally visualization tools that can be used analyze and display various aspects of the modeled traffic flow data such as shortest paths, travel time, flow compositions on a network in user-equilibrium is described.
Duane F. Marble, Zaiyong Gou, Lin Liu and James Saunders (***--)
Recent advances in the exploratory analysis of interregional flows in space and time.
Innovations in GIS 4, ed. Zarine Kemp.
The paper emphasizes a ‘researcher’s toolkit’ that supports viable exploratory analysis of spatial-temporal interaction databases. This toolkit should provide the researcher with an ability to easily adopt either an exploratory or confirmatory approach, as the specific situation requires.
The initial goal has been to develop a toolkit that significantly improves upon the existing methods of examining spatial pattern of interregional flows.
The visualization-based EDA tool described represents a substantial advancement over the previous technology available to support the scientific investigation of interregional flows. However, it still suffers from the severe limitation of being able to present only a limited number of O-D pairs simultaneously without degrading the visual display.
Edmond Mersobian et. al.
Mining Geophysical Data for Knowledge.
IEEE Expert, October 1996
A system called ‘Oasis’ (Open architecture scientific information system) for exploratory data mining (EDM) and analysis for scientific hypothesis testing is developed. In this article the authors explain how scientists can use this flexible, extensible computing environment for data analysis, knowledge discovery, visualization, and collaboration.
A good conceptual description of the system architecture about how it works is provided. The process is iterative and at each desired cycle the user can visualize result and interact with the process to improve the result.
Gennady L. Andrienko and Natalia V. Andrienko
Interactive maps for visual data exploration
Int. J. Geographic Information Science, 1999, Vol. 13, No. 4, 355-374
A good literature review on interactive manipulation of graphic display in exploratory data analysis is presented in this paper. A software system called ‘Descartes’ is designed to support visual exploration of spatially referenced data. Descartes offers two integrated services: automated presentation of data on maps, and facilities to interactively manipulate these maps. Descartes selects suitable presentation methods according to characteristics of the variables to be analyzed. Various interactive techniques for map manipulation are developed that could enhance the expressiveness of maps and thus promote data exploration. The interface is designed such that maps are not only visual but also can be manipulated.
Web (for Descartes): http://allanon.gmd.de/and/java/iris
Hing-Yan Lee and Hwee-Leng Ong
Visualization Support for Data Mining
IEEE Expert, October 1996, 69-75
The authors present a data mining software system called ‘WinViz’. WinViz is a data mining system that exploits the powerful combination of visual representation, interactivity, and machine-learning techniques. It uses a multidimensional visualization technique. A primary underlying design principle is "seeing your data in a single picture". WinViz’s visualization interface supports the display of multidimensional data and the visual formulation of queries interactively.
An inteface called Multidimensional Data Visualization (MDV) is developed. Display may work well but the result is not readily apparent to the user. Relationship among different variables is also difficult to understand.
Jiawei Han, K. Koperski and N. Stefanovic
GeoMiner: A System Prototype for Spatial Data Mining
Ref: ..................
A spatial data mining system prototype, GeoMiner, has been designed and developed based on the experience in the research and development of relational data mining system, DBMiner, and the research (of the authors) into spatial data mining. The data mining power of GeoMiner includes mining these kinds of rules: characteristics rules, comparison rules and association rules. Inclusion of mining classification rules and clustering rules is a future possibility.
The SAND (Spatial and Nonspatial Data) architecture is applied in the modeling of spatial databases, whereas GoeMiner includes the spatial data cube construction module, spatial on-line analytical processing (OLAP) module, and spatial data mining modules. A spatial data mining language GMQL is designed and implemented as an extension of Spatial SQL for spatial data mining. An interactive, user-friendly data mining interface is constructed and tools are implemented for visualization of discovered spatial knowledge.
Web: http://db.cs.sfu.ca/DBMiner
John H. Ganter (*****)
Display Techniques for Dynamic Network Data in Transportation GIS
1994, ref .............. ?
This paper discusses a prototype software system that uses computer-assisted visualization techniques to portray dynamic and detailed traffic data. It tries to answer what such a traffic visualization system look like and how might it be constructed.
There is a summary discussion on how people acquire and process spatial information at perceptual level. While providing a demo of the software, the underlying data and structures are discussed. It also tries to integrate GIS into the system.
The main topics of the paper are:
- Traffic data types and sources: Two main methods are discussed – induction loop and colsed-circuit television (CCTV). Laser interferometry is another technique mentioned. All these methods gather huge and accurate temporal and spatial data.
- Requirement for a prototype system: a ‘generic user’ should be able to I) filter, ii) symbolize and iii) control time. The software prototype is called Dynamic Network Display (DND) – the implementation strategy was to modify GIS technology (Arc/Info and AML) to avoid low-level database and graphics programming.
- Visual variables and time display: It uses seven variables defined by Bertin (1983) – these are position, size, value, texture, hue, orientation and shape. Time is displayed by animation.
Krzysztof Koperski, Junas Adhikary and Jiawei Han {koperski, adhikary, han}@cs.sfu.ca
Spatial Data Mining: Progress and Challenges Survey Paper.
1996 SIGMOD Workshop on Research Issues on Data Mining and Knowledge Discovery, Montreal, Canada, June.
This paper summarizes recent works on spatial data mining, from spatial data generalization, to spatial data clustering, mining spatial rules etc. It’s a good survey on various studies on data mining – focusing on ‘methods’ of spatial data mining.
Although nothing is written on visualization, this paper is useful as a background study. It covers many aspects of spatial data mining, such as – spatial data structures, spatial data mining architecture, spatial-data-dominant generalization, non-spatial-data-dominant generalization and various algorithms for SD and NSD generalization using CLARANS software.
In the future direction of study section, there are mention on visualization – developing a GUI for spatial data mining query language and multidimensional rule visualization.
Lars Bernard, Benno Schmidt, Ulrich Streit, and Christoph Uhlenkuken
Managing, Modeling, and Visualizing High-dimensional Spatio-temporal Data in an Integrated System.
GeoInformatica 2:1, pp. 59-77, 1998
An approach to in interoperable object-oriented GIS-framework for atmospheric modeling (AtmoGIS), which can be used to implement integrated information systems, is presented. The consideration of user-workflows leads to the specification of the projected system. Using an object-oriented approach, the system is based on a spatio-temporal database management system, a mesoscale model and an environment for scientific visualization.
Lin Liu, 1995. (*****)
Interactive Visualization and Analysis of State-to-state Migration Flows.
GIS-LIS ’95 Annual Conference Proceedings.
This paper presents a new methodology to explore state-to-state migration flow patterns in the United States. The proposed methodology integrates scientific visualization, exploratory data analysis and dynamic graphics to provide an interactive, graphic environment for exploring migration flow pattern. This methodology has been implemented in a prototype system on a Unix-based graphics workstation, written in C, Xlib and Motif. It offers four graphic views: a migration flow view, a choropleth view of origin and destination (O-D) states, a statistical view of migration, and, and a statistical view of socio-economic characteristics of states, to graphically represent migration data from different perspectives. A dynamic brushing technique has been implemented to link the four views; therefore revealing their relationships. These relationships provide the basis for generating migration flow patterns. The migration flow view is underlain graphically by the choropleth view that represents regional characteristics; thus displaying the relationship between migration and regional characteristics. The user can interactively select a set of states and examine the bi-directional migration flows among them. In addition, the content of the migration flow view can be adjusted by interactively setting an O-D distance range or a flow magnitude range.
Mark Gahegan (****-)
Four barriers to the development of effective exploratory visualization tools for the geosciences.
Int J. Geographic Information Science, 1999, Vol. 3, No. 4, 289-309
This paper outlines four specific problems that appear to represent considerable obstacle to the development of visualization strategies for use within the domain of geography and the Earth sciences. These are: 1) the speed of graphical rendering, 2) the management of perceptual anomalies due to visual combination effects, 3) the vast range of potential approaches and mappings (the complexity of the visual assignment process), and 4) the orientation of the user into an artificial or virtual reality. Each problem is discussed in terms of the visualization of geographical data for the purpose of exploratory visual analysis.
A variety of graphical languages are available as the basis for scene construction. Visualization environments ultimately map their scenegraph onto one of these languages. Three popular standards – OpenGL, VRML and DirectX/Direct3D – are briefly reviewed. Results of experiments with different visualization platforms show some major prblems with the speed of graphical rendering.
Web: http://www.cs.curtin/edu/au/gis/visualization/
Monica Wachowicz
GeoInsight: an approach for developing a knowledge construction process based on the integration of Gvis and KDD methods. (unpublished?)
http://www.geog.utah.edu/~hmiller/gkd_text/
This paper is a preliminary discussion on examining how the integration of Gvis and KDD introduces new challenges and conditions resulting from the complexity of the process of exploring data in space and time. From a system perspective, the issues are related to effectively support user-data interactions in both the underlying data models and high-level user interfaces. The main issue is to make the knowledge construction process very flexible and facilitate interactive exploration of spatio-temporal data. In this paper, the underlying principles and key developments of the past decade in the GVis and KDD research are reviewed. The review provides a base from which the commonality of goals and potential integration of Gvis and KDD is presented. Finally, the GeoInsight approach is proposed as a long-term strategy towards this integration.
Qin Tang (*****)
A Personal Visualization System for Visual Analysis of Area-based Spatial Data
GIS-LIS ’95 Annual Conference Proceedings, 767-776
This paper describes the design of a visualization system based on personal computers for exploring area-based spatial-temporal data. Conceptual discussion on combining visualization and spatial data analysis is presented. The system features, requirements, components and structure are discussed. The research illustrates how to develop a visualization system through enhancing traditional cartographic methods for visual spatial data analysis.
It proposes the Visual Data Analysis (VDA) method which is a combination of computer graphics and EDA. VDA explicitly links visualization and data analysis.
The point of interest for this system is that it runs on a PC 386/486 and requires a VGA monitor. The operations are mouse-driven (keyboard is also acceptable). The interface consists of four windows – of those one is used for interactive analysis.
S. Shekhar, C. Lu, X. Tan, S. Chawla, R. R. Vatsavai
Map Cube: A Visualization Tool for Spatial Data Warehouses
http://www.geog.utah.edu/~hmiller/gkd_text/
A map is the core output of a map cube. Issues involved in visualization of geographic features are discussed, such as, how to represent features with different level of measurements. Data cube operator is used to generate the union of a set of alpha-numeric summary tables corresponding to a given aggregation hierarchy. The concept of data cube is extended to spatial domain by ‘map cube’, an operator which takes the base map, base table, cartographic preference, etc. and generates an album of maps.
Web: http://www.cs.umn.edu/Research/shashi-group
Steven L Salzberg (salzberg@cs.jhu.edu)
On Comparing Classifiers: Pitfalls to Avoid and a Recommended Approach.
Data Mining and Knowledge Discovery, ed. Usama Fayyad, pp. 317-327, 1997
An important component of many data mining projects is finding a good classification algorithm, a process that requires very careful thought about experimental design. If not done very carefully, comparative studies of classification and other types of algorithms can easily result in statistically invalid conclusions. This is especially true when one is using data mining techniques to analyze very large databases, which inevitably contain some statistically unlikely data. This paper describes several phenomena that can, if ignored, invalidate an experimental comparison. These phenomena and the conclusions that follow apply not only to classification, but also to computational experiments in almost any aspect of data mining. The paper also discusses why comparative analysis is more important in evaluating some types of algorithms than for others, and provides some suggestions about hot to avoid the pitfalls suffered by many experimental studies.
Usama M. Fayyad, Piatetsky-Shapiro and Padhraic Smyth
From Data Mining to Knowledge Discovery: an Overview
(Data Mining book – ref.............?)
This paper is a good overview of different aspects of data mining. This chapter of the book presents an overview of the state of the art in this field. First, the relation between data mining and knowledge discovery is clarified. Definitions of data mining and KDD are given in the beginning. Then application issues including guidelines for selecting an application and current challenges are covered. The discussion relates methods and problems to applicable chapters in the book, with the goal of providing a unifying vision of the common overall goals shared by the chapters.