Three Dimensional Visualization of the World Wide Web
Steve Benford, Ian Taylor, David Brailsford, Boriana Koleva, Mike Craven, Mike Fraser, Gail Reynard , Chris Greenhalgh
University of Nottingham
Nottingham, UK
Email: ac.uk
Introduction
Although large-scale public hypermedia structures such as the World Wide Web are popularly referred to as "cyberspace", the extent to which they constitute a space in the everyday sense of the word is questionable. This paper reviews recent work in the area of three dimensional (3D) visualization of the Web that has attempted to depict it in the form of a recognizable space; in other words, as a navigable landscape that may be visibly populated by its users. Our review begins by introducing a range of visualizations that address different aspects of using the Web. These include visualizations of Web structure, especially of links, that act as 3D maps; browsing history; searches; evolution of the Web; and
the presence and activities of multiple users. We then summarize the different techniques that are employed by these visualizations. We conclude with a discussion of key challenges for the future.
web下载官方下载Visualizing the structure of the Web
Various visualizations have been developed to show the structure of a region of the Web. These are intended to provide users with 3D maps to guide browsing. Common assumptions behind these visualizations are that 3D interfaces may be capable of displaying a larger volume of information in a single display than 2D, and that users may more easily learn the structure of the Web as a result of flying though a 3D landscape [Benford 1996]. Perhaps the most obvious representation of Web structure is as a graph of nodes connected by links. Natto [Shiozawa 1997] visualizes a number of Web pages by drawing such a node/link graph that is initially distributed on a flat horizontal plane. The placement of the nodes is dictated by attributes of the Web page (e.g. its size, title, number of images) which are mapped to the two-axis of the plane. The user may select nodes and raise them vertically to de-occlude the structure. Adjacent (linked) nodes maintain a close proximity to the raised nodes so that the structure is gradually "disentangled" from the plane. Figure 1 shows a typical result of this process. Natto demonstrates a method of de-occlusion by manual control, however, there are limits to the number of nodes that may comfortably occupy the flat plane before selection becomes difficult. Als
o, the range of pages is fixed when the visualization is generated.
__________________________
FINAL DRAFT of paper which appeared in ACM "Computing Surveys" 31(4es), December 1999.
Figure 1: A selection of pages in Natto.
Narcissus [Hendley 1995] also produces a graph-like representation. Occlusion is automatically reduced through the metaphor of attractive and repulsive forces. Web pages (nodes) exert repulsive forces on each other, whereas the links between them lead to attractive forces. Simulating these forces results in tightly inter-linked sets of pages being grouped into spatial clusters within the visualization. Figure 2 demonstrates a resulting graph.
Figure 2: Web site visualisation with clustering in Narcissus.
Although occlusion can still occur, the problem of two objects occupying the same space at the same time is eliminated. An additional scaling mechanism is offered by agglomerating clusters into a single, i
dentifiable object. Narcissus does this by surrounding clusters with a translucent surface to effectively remove most of the detail.
An alternative approach has been demonstrated by Munzner and Burchard [Munzner 1995] in the construction of three-dimensional hyperbolic space that supports "focus + context" exploration of hypertext structure. The hyperbolic space is formed inside a sphere whose surface is conceptually an infinite distance from the centre. Nodes close to the centre of the sphere appear large and it is here that the focus is achieved. Figure 3 provides an example of this layout for several WWW pages.
This work is similar to that of Lamping et al. Lamping 1995] but adds a third dimension with the intention of allowing a higher information density within the space.
Figure 3: Hyperbolic layout of a number of pages.
In the HotSauce [HotSauce 1999] Web browser from Apple Computers the links are implicit. The HotSauce browser represents Web pages as floating text labels in a 3D space. Pages at the higher level of the hierarchy in a Web server appear large in the foreground whilst the next level would be further away, distributed evenly in the background of their parent page label. The user can "walk" through the labels, progressively bringing lower levels into the foreground. Labels can be repositioned or selected to invoke the content. Figure 4 contains an initial HotSauce view of the Apple Computer Web site.
The Perspecta SmartContent [SmartContent 1999] viewer adopts a similar approach to HotSauce with the addition of link representation. It also allows the user to move the viewpoint into the page hierarchy to reveal detail at lower levels.
Figure 4: Pages as viewed in the HotSauce Browser
Two further visualizations of Web structure move away from a graph like representation.
The Open Text Web Index visualization [Bray 1996] visualizes several variables associated with Web sites. Parameters such as number of pages, number of links to and from these pages, and domain identifiers, define a composite object that is placed in a 3D space. Such objects combine distinct visual components to represent information around a single unit, i.e. the Web site. The distance between them is determined by the degree of connectivity between the two sites. A view of these sites is shown in figure 5. The lack of detail in the visualization prohibits its use as a page browser but acts as a map at the site level.
Figure 5: A view of the web generated from the Open Text Web index.
MAPA [Durand 1998] is a visualization that aims to improve navigation in large Web sites of between 500 and 50,000 pages. MAPA presents pages as square icons that stand in rows and columns on a flat plane. A focus page is placed at the front edge of the plane and its child pages form a row behind the focus page. Each child page then has its children behind it so as to form a single column behind the page. Essentially, this aims to provide a layout mechanism with minimal visual clutter. Figure 6 demonstrates this layout with a view of the pages within a company Web site.
Figure 6: A Web site viewed in MAPA.
Visualizing browsing history
Although they share many common techniques with the systems described above, visualizations of browsing history provide images of where the user has been as opposed to maps of where they might go.
WebPath [FrÈcon 1998] generates a three-dimensional representation of a Web browsing history within the DIVE distributed virtual reality environment. WebPath runs alongside a conventional browser and is updated whenever a page is visited. Each page is represented as a cube upon which one of three page properties is displayed (the page background image, the first image in-lined in the HTML or the page background colour). The vertical position of the cube is incremented at each visit so as that the most recently visited cube is at the highest point. Semi-transparent layers divide the space so that the older pages gradually become more obscure due to a fogging effect. This effect can be seen over a number of successive layers in figure 7.

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。