Graham John Wills

Navigating large networks with hierarchies

This paper is aimed at the exploratory visualization of networks where there is a strength or weight associated with each link, and makes use of any hierarchy present on the nodes to aid the investigation of large networks. It describes a method of placing nodes on the plane that gives meaning to their relative positions. The paper discusses how linking and interaction principles aid the user in the exploration. Two examples are given; one of electronic mail communication over eight months within a department, another concerned with changes to a large section of a computer program.<<ETX>>

NicheWorks—Interactive Visualization of Very Large Graphs

Abstract The difference between displaying networks with 100–1,000 nodes and displaying ones with 10,000–100,000 nodes is not merely quantitative, it is qualitative. Layout algorithms suitable for the former are too slow for the latter, requiring new algorithms or modified (often relaxed) versions of existing algorithms to be invented. The density of nodes and edges displayed per inch of screen real estate requires special visual techniques to filter the graphs and focus attention. Compounding the problem is that large real-life networks are often weighted graphs and usually have additional data associated with the nodes and edges. A system for investigating and exploring such large, complex datasets needs to be able to display both graph structure and node and edge attributes so that patterns and information hidden in the data can be seen. In this article we describe a tool that addresses these needs, the NicheWorks tool. We describe and comment on the available layout algorithms and the linked views int...

High interaction graphics

Abstract Examining data using graphical tools, such as histograms, quantile plots, scatterplots and the like, is a necessary part of any serious analysis effort. With the advent of inexpensive graphics-capable desktop computing, such tools are generally available. But the use of computers enables more than simply reproducing static plots on a display; it allows users to interact with plots, changing parameters, querying, zooming and linking plots together so that interesting features of one plot can be seen in the light of the others. In this paper we discuss the core features of interactive graphics, investigate how familiar plots can be made interactive and show examples of interactive graphics for general and specific data analysis.

Selection: 524,288 ways to say "this is interesting"

Visualization is a critical technology for understanding complex, data-rich systems. Effective visualizations make important features of the data immediately recognizable and enable the user to discover interesting and useful results by highlighting patterns. A key element of such systems is the ability to interact with displays of data by selecting a subset for further investigation. This operation is needed for use in linked-views systems and in drill-down analysis. It is a common manipulation in many other systems. It is as ubiquitous as selecting icons in a desktop GUI. It is therefore surprising to note that little research has been done on how selection can be implemented. This paper addresses this omission, presenting a taxonomy for selection mechanisms and discussing the interactions between branches of the taxonomy. Our suggestion of 524,288 possible systems [2/sup 16/ operation systems/spl times/2 (memory/memoryless)/spl times/2 (data-dependent/independent)/spl times/2 (brush/lasso)] is more in fun than serious, as within the taxonomy there are many different choices that can be made. This framework is the result of considering both the current state of the art and historical antecedents.

An interactive view for hierarchical clustering

The paper describes a visualization of a general hierarchical clustering algorithm that allows the user to manipulate the number of classes produced by the clustering method without requiring a radical re-drawing of the clustering tree. The visual method used, a space filling recursive division of a rectangular area, keeps the items under consideration at the same screen position, even while the number of classes is under interactive control. As well as presenting a compact representation of the clustering with different cluster numbers, this method is particularly useful in a linked views environment where additional information can be added to a display to encode other information, without this added level of detail being perturbed when changes are made to the number of clusters.

Visualizing Reference Databases

This paper describes a graphical method for visualizing reference database searches. The motivation for inventing this technique comes from analyzing the Current Index of Statistics reference database. This database contains 128 thousand references to articles from statistical journals, conference proceedings, and books, published during the last 20 years. The paper traces the evolution of the bootstrap technique, a statistical research breakthrough, in the statistical literature, shows yearly trends, discovers which journal publish articles on bootstrapping, and identifies books on this subject.