Populated Information Terrains

This page contains information on some of the visualization techniques being employed by the Communications Research Group at the University of Nottingham.

We refer to the generated visualizations as Populated Information Terrains (PITS) to reflect the the fact that a key feature of this work is that these visualizations are intended to be multi-user applications in which the users are explicity embodied in the virtual environment and are thus visible to each other.

Techniques for Generating the Visualisations

We have identified four approaches to generating the visualizations, each applicable to different types of data:

Named after Michael Benedikt, this approach directly maps fields from data objects into extrinsic and intrinsic dimensions of the displayed objects. In this approach the user generates a schema which contains the desired mappings from the fields of the data objects to the extrinsic and intrinsic dimensions - an example might be to map the log of a person's age to the x-ordinate, and map a persons job type to the shape of the displayed object.
Statistical methods have been used to analyse collections of data (often documents) in an attempt to group objects together according to some measure of semantic "closeness" (i.e. do they logically belong together). The resulting proximity measures are typically scaled and returned as numerical values that are then used to cluster the objects in a data space. Systems adopting this and similar approaches include VIBE and BEAD (both for visualising collections of documents).
Hyper-structures: Some databases support the notion of explicit relations or links between objects (e.g. schema based on entity-relationship models or hyper-media and Network Information Retrieval systems such. Gopher, ARCHIE, WAIS and World Wide Web). The resulting structures might be visualised by applying three dimensional graph drawing techniques. In turn these might be extended through visualisation techniques such as fish-eye views, cone-trees and perspective walls.
This final approach relies on users or system implementors designing appropriate visualisations. The first approach is to use real-world metaphors such as a fly-through library interface for a document store; cities, buildings and rooms for organisational information in Directory services; and maps of the physical world for geographical information. A second approach is to allow humans to construct and organise the information terrain themselves on an ad-hoc basis (effectively how files are organised under the desktop metaphor). This approach might also extend previous work on rooms metaphors in user interfaces.

Example Visualisations

Statistical methods

An example of this approach is VR-VIBE, developed at Nottingham. This extends the VIBE approach into three dimensions and creates visualizations of bibliographies. Users specify keywords that they wish to use to generate the visualization and place these keywords in 3D-space. Representations of the documents are then displayed in the space according to how relevant each document is to each of the keywords (this relevance is computed by searching the documents for the keywords specified by the user and recording the number of matches). The position of a document depends on the relative importance of each of the keywords to it; thus a document equally spaced between two keywords is equally relevant to both, while a document close to a particular keyword is relevant to that keyword only. Absolute relevance is depicted by the colour of the documents representation, the more relevance the document, the brighter the colour - this is necessary to distinguish between documents that are equally only slightly relevevant from documents which are highly relevant but are placed in the position because the proportion of their relevance to each keyword is the same.

The following image shows an example VR-VIBE visualization of a CSCW bibliograpy containing 1581 entries and five keywords, also visible are some other users browsing the information space. The position of the keywords are marked by octahedrons, with the text drawn alongside. Only documents with an absolute relevance greater than specified threshold are displayed allowing the user to filter out less relevant documents. A 3D scrollbar, visible to the left of figure 2 allows the values of the relevance threshold to be varied.

The screenshot shows a pyramid of blocks, each representing an entry in the bibliography, and users navigating through it.

This image shows the same visualization from a different viewpoint. More information about VR-VIBE and some more screen shots of it in use are available.

A view of the same representation from a different angle.

Hyper-structures

This image shows a visualization of part of a UNIX filestore as a cone-tree, illustrating the hyper-structure approach.

The figure shows links spreading out from a central point in a cone, with the end of each link also forming the basis of a cone.

Another approach that we are using to visualizer hyper-structures is known as "Force Directed Placement". In this approach links between nodes are treated as spring forces. The nodes are initially placed either randomly or with some nodes in specified locations and then the forces between nodes are simulated. This has the effect or bringing connected nodes into proximity with one another. A repulsive force between nodes ensures that nodes do not become so close as to make the visualization too dense.

This image shows a visualization of structure with 213 nodes.

The figure shows a three-dimensional network of nodes and links with a seemingly random structure.

This image demonstrates the sort of visualization produced when a hierarchical structure is visualized

The figure shows a more structured three-dimensional network of nodes and links.

Populated Information Terrains

Contents

Techniques for Generating the Visualisations

Example Visualisations

Statistical methods

Hyper-structures

Links

Contributing Technologies

Publications