SFDC 1007
`SFDC 1007
`
`
`
`
`
`
`
`
`
`

`
`A Zooming Web Browser
`
`Benjamin B. Bederson, James D. Hollan,
`Jason Stewart, David Rogers, Allison Druin, David Vick
`Computer Science Department
`University of New Mexico
`Albuquerque, NM 87131
`{bederson, hollan, jasons, drogers, allisond, dvick}@cs.unm.edu
`http://www.cs.unm.edu/pad++
`
`ABSTRACT
`The World Wide Web (WWW) is becoming increasingly important for business, education, and entertainment. Popular web
`browsers make access to Internet information resources relatively easy for novice users. Simply by clicking on a link, a new
`page of information replaces the current one on the screen. Unfortunately however, after following a number of links, people
`can have difficulty remembering where they’ve been and navigating links they have followed. As one’s collection of web pages
`grows and as more information of interest populates the web, effective navigation becomes an issue of fundamental importance.
`
`We are developing a prototype zooming browser to explore alternative mechanisms for navigating the WWW. Instead of having
`a single page visible at a time, multiple pages and the links between them are depicted on a large zoomable information surface.
`Pages are scaled so that the page in focus is clearly readable with connected pages shown at smaller scales to provide context.
`As a link is followed the new page becomes the focus and existing pages are dynamically repositioned and scaled. Layout
`changes are animated so that the focus page moves smoothly to the center of the display surface while contextual information
`provided by linked pages scales down.
`
`While our browser supports multiscale representations of existing HTML pages, we have also extended HTML to support multi-
`scale layout within a page. This extension, Multi-Scale Markup Language (MSML), is at an early stage of development. It cur-
`rently supports inclusion within a page of variable-sized dynamic objects, graphics, and other interface mechanisms from our
`underlying Pad++ substrate. This provides sophisticated client-side interactions, permits annotations to be added to pages, and
`allows page constituents to be used as independent graphical objects.
`
`In this paper, we describe our prototype web browser and authoring facilities. We show how simple extensions to HTML can
`support sophisticated client-side interactions. Finally, we discuss the results of preliminary user-interface testing and evaluation.
`
`Keywords: world-wide web, browser, information navigation, zooming, information visualization, multiscale information, ani-
`mated user interface, Pad++.
`
`1. INTRODUCTION
`In 1945 Vannevar Bush [8] envisioned “a future device for individual use, which is a sort of mechanized private file and library.”
`He termed this device a memex and proposed a form of associative indexing in which arbitrary pieces of information could be
`linked together such that “when one of these items is in view, the other can be instantly recalled by tapping a button.” He further
`conjectured that “wholly new forms of encyclopedias will appear, ready made with a mesh of associative trails running through
`them, ready to be dropped into the memex and there amplified.” Today, fifty years later, we have the World Wide Web and a
`memex in the form of web browsers. See [4] for an overview of the WWW.
`
`The increasing number of users and the ever-growing quantity of information available on the web present challenging interface
`and navigation problems. There are a variety of human factors [19] issues that need to be addressed. A larger number of users
`means that people with diverse talents, interests, and experiences will be on-line via the web. Many will be novices with little
`prior experience with computers. A simple click of the mouse can bring a user from their friend’s home page to unknown desti-
`nations across the world. Traditionally, following a cross reference meant shuffling across the library to find another volume.
`While time-consuming, this reinforced the transition that was taking place. The difficulties that novice users confront can be
`
`SFDC 1007
`
`

`
`instructional to developers. While experts may not have as much difficulty, they experience the same cognitive burdens, and
`may just have a higher threshold before they experience similar difficulties.
`
`While the immediacy of traversing information links offers many advantages, it can also make it difficult to maintain an intuitive
`sense of where one is, and how one got there - leading to the frequently described sense of being lost. This is a classic problem
`of hypertext systems. Part of the problem can be attributed to windows-based interfaces. Current window systems don’t readily
`support showing more than a few pages at a time. In addition, each page is usually in a separate window with no depiction of
`relationships to other windows. Popular WWW browsers, like other applications built according to current tiled or overlapping
`windows philosophies, also have this same problem, although they do offer limited methods to aid navigation by keeping track
`of interesting sites - usually in hierarchical sets of hotlists or bookmarks.
`
`Several groups have proposed alternatives and extensions to browsers to address some aspects of this problem. Oostendorp
`describes the PAINT system (Personalized Adaptive Internet Navigation Tool) [25]. It provides an interface for accessing hier-
`archies of bookmarks in a style similar to the NEXTStep interface. WebMap is a browser extension that shows a graphical rela-
`tionship between web pages [11]. Each page is represented by a small circle that can be selected to display the actual page. The
`links between pages are colored to indicate information about the links, such as whether it is a link to a different server or
`whether the destination page has already been read. These graphs may be saved and used by others.
`
`While the web is inherently cyclic, it is easier to visualize hierarchies, and so many web visualizations are based on hierarchies
`extracted from the graph of the web. Some interesting work focuses on alternative visualizations [24]. Furnas [14] shows how
`multitrees can be used to represent a collection of hierarchies sharing parts of the underlying data. One application of multitrees
`is visualization of bookmarks from multiple individuals[34]. Furnas [16] also describes a framework for characterizing how dif-
`ferent structures influence effective view traversal, the mechanical process of moving between information items, and view nav-
`igation, finding good paths to information items.
`
`Another approach to visualizing large information spaces that can be applied to web browsing and navigation involves tech-
`niques to show detail at particular nodes while maintaining context. One general approach, fisheye views [13], has been
`extended with graphics [30], three dimensions [9][22], hyperbolic representations[20], animation [10], and zooming [2][3][28].
`Other techniques include exploiting a large virtual space [12], using lenses or filters [5][23][31], and visualizing two dimensional
`layouts [1][21].
`
`In addition to the difficulty of finding information, it becomes ever more important to tailor information for one’s own needs.
`Also rather than searching oneself can be sensible to go by other’s recommendations. This is the basis for commercial services
`such as Yahoo [36], and follows the often effective strategy of exploiting recommendations from those one knows and trusts
`[35].
`
`Annotations are another important information tailoring facility. Annotations are personal markings that can be used to highlight
`and comment on information for oneself and others. One interesting approach to annotation on the web separates the annotations
`from the original documents and stores them in a special annotation server [29]. Used with an enhanced browser, displaying a
`new page automatically brings in the annotations of others and integrates them into the page.
`
`In the sections that follow, we describe our zooming web browser and the attempt to use animation and multiscale representation
`of context to support more effective web navigation. In addition to visualization of standard HTML pages, we introduce exten-
`sions to HTML that allow more sophisticated presentations and client-side interactions. We demonstrate the beginnings of direct
`manipulation graphical authoring tools and show how annotation can be supported as a form of authoring. Finally, we present
`the results of initial user testing and envision a scenario for future web use in the classroom.
`
`2. A ZOOMING WEB BROWSER
`Navigating the WWW presents a struggle between focus and context. As one browses or searches the web the need for detailed
`views of specific items conflicts with the need to maintain a global view of context and history of traversal. This struggle is
`made more difficult by the haphazard organization of the WWW. Information items closely related by links are not necessarily
`closely related by content nor in terms of the user’s information needs. At times, one seems more likely to find something of
`interest when not looking for it than when specifically searching. This serendipitous contact with information, though at times
`frustrating, can also be an advantage. The challenge is how to best support both incidental and intentional access while organiz-
`
`

`
`ing useful information so that it can be effectively retrieved again in the future.
`
`We are exploring dynamic multiscale techniques to support focus and context during navigation of large information spaces. To
`accomplish this we are building a zoomable web browser using Pad++, a substrate for building multiscale dynamic user inter-
`faces [2][3][27][28]. Pad++ provides an extensive graphical workspace where dynamic objects can be placed at any position
`and at any scale. Pad++ supports panning and zooming. Zooming can involve simple geometric scaling or what we term seman-
`tic zooming, in which rendering of objects can vary based on factors in addition to scale, such as context of the task or complex-
`ity of the information being displayed. Pad++ is built as a widget for Tcl/Tk, a scripting language and user-interface library
`[26][33].
`
`Pad++ allows WWW pages to remain visible at varying scales while they are not specifically being visited, so the viewer may
`examine many pages at once. In addition, Pad++ allows the user to zoom in and out of pages, enabling explicit control of how
`much context is viewed at any time. To orient themselves, users can simply zoom back to view a number of web pages. To get
`more detailed views of a particular page they can zoom in. We think this variable scale contextual display of web pages can pro-
`vide important support for navigation. We are currently exploring a tree layout system that permits users to dynamically add to
`and reorganize a tree of web pages. Using our Pad++ web browser, users navigate a space filled with familiar objects, not iconi-
`fied representations of those objects.
`
`Our dynamic Pad++ tree browser combines a basic focus-driven layout with automatic zooming and panning to support naviga-
`tion. The software allows the user to select a focus page. That selection animates the page to occupy a larger section of the dis-
`play. Pages farther from the focus page get increasingly smaller, resulting in a graphical fisheye view [30]. See Figures 1 and 2
`for snapshots of the Pad++ web browser during reorganization.
`
`Figure 1: Snapshot of Pad++ Web Browser.
`
`Figure 2: Another view of same web pages.
`
`The Pad++ WWW browser combines Pad++’s interactive multiscale display with dynamic objects that can restructure them-
`selves in response to user actions. Clicking on a link brings up a new page, adds it to the tree of pages, and causes the tree to
`restructure itself. Unlike other web browsers that immediately replace the current page with a new page, the restructuring pro-
`cess is animated so that users can understand how the tree is being reorganized. The animation helps maintain object constancy
`and the graphical depiction of links highlights relationships between pages. The new page becomes the current focus and is
`moved to the center of the screen, at a size suited for viewing. The user may designate any existing page to be the current focus
`by clicking on it.
`
`As in earlier fisheye displays, our basic layout function assigns a degree-of-interest value to each node in the tree based on its
`distance from the focus page. We define the distance to be the shortest path between two pages[13]. This value is then used to
`
`

`
`determine the size of each node. See [11] for a description of other hierarchical layout techniques not based on fisheye views.
`
`The layout described above provides a sense of context while following links. We have also implemented an alternative camera
`mode of navigation. It shows the web of links on one side of the screen with a zoomed in view of the focus page on the other
`side of the screen. A camera is depicted along with the web of pages. The camera can be dragged around or automatically ani-
`mated through the web. The zoomed in view shows the page the camera is currently looking at (Figure 3). This mode also sup-
`ports automated tours. For example, one type of camera can take you on a tour of all the parts of your saved web pages that have
`changed since you last looked at them.
`
`We are currently experimenting with more flexible mechanisms for dynamic tree layout and interaction. These include exploring
`alternative visualizations and better methods for managing and interacting with large dynamic trees. New tree layout methods
`will work with any kind of item on the Pad++ surface. Thus, in addition to HTML pages, users will be able to create spaces
`using any Pad++ object, including drawings, interactive maps, and text.
`
`Figure 3: Camera view of web pages.
`
`The new layout code is designed to be hierarchical, so that users may designate subtrees to have different layouts. This allows
`greater freedom in grouping and display. For example, a certain information tree may contain nodes with subtrees consisting of
`hundreds or thousands of nodes each. These nodes could exploit a hyperbolic layout to compress the information and the hyper-
`bolic nodes themselves might be layed out radially [20].
`
`Another topic we are exploring involves tradeoffs between maintaining pointers to information on the Web and making copies of
`the information locally. For example, a user might want to copy items from a remote page to prevent that information from
`being lost. At other times users may wish to maintain only pointers to information since it is being maintained elsewhere. There
`are interesting related issues of annotation, that we discuss later, as well as issues of maintaining annotation placement as pages
`change.
`
`

`
`3. AUTHORING WEB PAGES
`Thus far, we have discussed visualization and navigation of WWW pages written using standard HTML. We are also exploring
`extensions to HTML to allow web page authors access to Pad++ multiscale visualization and layout facilities. This extension,
`Multi-Scale Markup Language (MSML), enables users to include arbitrary Pad++ objects in web pages. We have implemented
`MSML using the HTML <Meta> tag and thus added features are invisible to HTML browsers.
`
`Important motivations for MSML are to make WWW pages and their components first-class Pad++ objects and allow authors
`using MSML to exploit all Pad++ facilities. In addition to adding zooming and other dynamic features to web pages, the goal is
`permit authors to manipulate and interact with any web page element. Making elements first-class Pad++ objects will result in
`authors being able to move, scale, delete, add, or modify them. Our approach is similar in spirit to the SELF project [32]. While
`much remains to be accomplished to support the full informational physics [3] we envision, all of the examples detailed below
`work in the current Pad++ Web browser.
`
`HTML provides very few basic types. Examples include text, images, bullets items, and horizontal rules. MSML in contrast,
`supports not only HTML types but also provides access to all the graphical features of Pad++. This gives users a richer toolkit of
`objects to use when creating documents: text (Postscript Type 1 fonts), lines, rectangles, ovals, lenses (providing filtering and
`alternative representation), portals (furnishing additional zoomable views of the Pad++ surface), compound objects created from
`these basic elements, and access to Pad++’s dynamic zooming and panning facilities.
`
`Zooming and Scale
`HTML provides header tags, <H1>, <H2>, etc., to indicate degree of importance of sections in WWW documents. However, this
`mechanism only works with text, not with other media such as images. MSML introduces a method to control the size of all
`types of objects, including images and graphics. We have used this, for example, to create a multiscale hierarchical outline (see
`Figure 4). All MSML extensions are written using special Meta-tag keys. This approach makes it clear that the included code is
`an extension and allows it to simply be ignored by standard HTML-based web browsers.
`
`A portion of the MSML required to create the outline depicted in Figure 4 is given below. The pad_scale key takes a single
`argument that multiplies the current scale and affects the size of all future objects until a /pad_scale key is seen.
`
`<html>
`Introduction
`
`<meta pad_scale=0.25>
`<ul>
`<li>Sistine Chapel
`<li>Push the interface metaphor
`<li>History
`
`<meta pad_scale=0.25>
`<ul>
`<li>Ivan Sutherland, SketchPad, 1963
`<li>William Donelson, MIT, 1978
`<li>George Furnas, Fisheye Views - Bellcore, 1986
`<li>Ken Perlin, David Fox, PAD - NYU, 1993
`</ul>
`<meta /pad_scale>
`
`</ul>
`<meta /pad_scale>
`...
`
`Another example use of MSML involves inclusion of a multiscale state map on a web home page. As the view is zoomed in,
`first counties, then cities, and then street maps of cities are shown. Finally, even the location of one’s home or work could be
`indicated on the street map. It would even be possible to continue zooming until a floor plan of home or work location becomes
`visible. See Figures 5-7 for a sequence of snapshots as we zoom into New Mexico. We first see county names and ultimately an
`Albuquerque street map.
`
`

`
`Below is the MSML code that produced the examples in Figure 5-7:
`
`<html>
`This is the New Mexico Map page
`<hr>
`<meta pad_tcl={msml_load_tcl http://www.cs.unm.edu/~bederson/pad/county.tcl county}>
`The map data is stored in a separate code file and is loaded using the pad_tcl tag. It passes a Tcl script that uses the MSML
`library function msml_load_tcl. This function takes two arguments: a URL to a Pad++ Tcl script and a tag name to be associated
`with every object the script creates. It is via this tag that objects are associated with the HTML page.
`
`Figure 4: Outline using MSML Scale Tag.
`
`Figure 5: New Mexico county data.
`
`Figure 6: NM zoomed into county names.
`
`Figure 7: NM zoomed into Albuquerque.
`
`The URL specifying county.tcl contains Tcl code that creates the county data. The msml_load_tcl function inserts it
`
`

`
`into the Web page and properly scales it. One advantage of MSML is that almost any object on the Pad++ surface can be used as
`an anchor, not just text and images. So once a user has zoomed into the New Mexico state map and located Albuquerque, the
`homes of the members of the Pad++ group as well as the Computer Science department itself could be anchors to other WWW
`pages, or to other points on the map, such as those of collaborators from other parts of the world.
`
`Interaction
`Traditionally, browsers come with pre-defined functions and all interactions with a web document are constrained to those func-
`tional abilities. Limited animation is possible through techniques such as server-push and client-pull. Forms, a simple interface
`built into most clients to collect information and send it to the server for processing, provide constrained GUI-like interactions.
`However, until recently, nothing supporting more interactive and flexible interfaces has been available.
`
`Currently there are a number of efforts to create more interactive WWW documents. The primary approach is to write code in a
`programming language, instead of HTML, that can be downloaded into a browser equipped to interpret the language. Sun’s Hot-
`Java project uses the Java language [18], Cygnus Support’s GNU Remote Operations Web (GROW) proposes to use GNU’s
`Guile extension language [17], and Microsoft’s Blackbird will use dynamically loadable object files [6]. By providing the ability
`to download and run code locally allows complicated animations, for example, to be encoded in a concise and network efficient
`manner.
`
`The Pad++ WWW browser contains a full Tcl interpreter. MSML provides mechanisms to include Tcl code within the page,
`instruct the browser to download either scripts or saved Pad++ data files over the WWW, and pass the code to the Tcl interpreter.
`This, of course, carries with it enormous security risks that we have not yet addressed. In the future we may restrict ourselves to
`a safe subset of the language as with Java or GROW. In our case this will initially be SafeTcl [7] but may also include support
`for other languages (e.g. Java).
`
`Figure 8: The grasshopper filter applied to the
`New Mexico County data.
`
`Figure 9: A WWW page containing Pad++ data
`files with some elements dragged out.
`
`Lenses
`Pad++ lenses can change the way objects look on the Pad++ surface [2][5][31]. In MSML, they can be particularly useful. For
`example, the data used to create the map of New Mexico consists of about 50K bytes of vertex data just to define the county
`boundaries and the city locations. There are many different kinds of data that one could be interested in displaying in relation-
`ship to a map of New Mexico (e.g., demographics of the local populations, geographical features of interest, etc.) If one were
`interested in not only providing an accurate street map to the Computer Science Department but also, say, an accurate count of
`the 1995 grasshopper population across the state, it would be pointless to include the geographical vertices twice. Instead, the
`
`

`
`same map can be used, and lenses supplied such that when viewed through the appropriate lens only the information of interest
`is visible. Figure 8 demonstrates the use of such a lens.
`
`Annotations and Authoring
`Because MSML allows users to interact with all elements of a WWW page, it provides a unique opportunity to explore annotat-
`ing and authoring of WWW pages. Based on experiences working with children, we have come to appreciate the need for sim-
`ple and intuitive ways to author WWW pages. One very natural authoring mechanism is to directly use and modify existing
`components of others peoples’ web pages. Every web page then becomes a potential supplier of components.
`
`We have begun to implement a drag-and-drop interface for authoring WWW documents. Figure 9 shows a page containing sev-
`eral Pad++ drawings. Just under that page, a composite figure was created by dragging elements out of the top web page (a kit of
`components constructed for young users to author zoomable hypertext stories) and putting them together. Our approach is to
`allow creation of web pages using the same direct manipulation techniques. In addition, we expect to provide layout support to
`facilitate creation of aesthetically effective web pages.
`
`A direct extension of this authoring technique allows users to maintain local copies of WWW pages with added annotations. A
`goal, much in keeping with the kind of personal information environment Vannevar Bush envisioned, is to enable users to create,
`save, and share annotated databases of WWW pages. This is similar to graphical hotlists but with the important difference that
`with MSML, users can maintain not only the HTML data for a particular page, but also associate arbitrary Pad++-based annota-
`tions. Pad++ supports a variety of annotations: not only can text be added, but entire WWW pages can be added to other WWW
`pages creating hierarchical meta-pages; graphics can be added, for example, to indicate an interesting section of a particular
`page, even when that page is scaled very small. Our goal is to provide rich annotation facilities by combining Pad++’s dynamic
`multiscale annotation ability with a control system that supports creation of virtual documents in which documents and annota-
`tions are separately controlled and maintained. See [29] for an example of one such control mechanism.
`
`To support effective views of collected web pages it is important to be able to show modifications over time. If one cares about
`a particular set of pages enough to include them in a personal collection it is likely that notifications of modifications will also be
`of value. We are exploring mechanisms to highlight changes. One technique highlights, by changing color or bounding with a
`rectangle, sections changed since some specific date (for example, the last time you viewed the document). Another uses scaling
`to show the history of changes. Older versions are shown at increasingly smaller scale. Marking changes is similar in spirit to
`efforts conducted by [35] using the WebWatch program. Unlike that effort we conjecture that one might not merely want to
`know that a document has been altered but also the details of the changes.
`
`4. USER TESTING RESULTS
`In November of 1995, we completed a pilot test of our zooming web browser. There were 14 test participants from the Univer-
`sity of New Mexico community, equally split between the College of Education and the Computer Science Department. A
`majority used email regularly, but almost no one used the WWW as frequently. Over a third of the participants from the College
`of Education had never used the WWW. All of the Computer Science participants had used the WWW. Before the pilot test
`began, a 5-minute demonstration of our browser was shown to participants. Participants were then asked to use the browser for
`30-60 minutes and subsequently completed a survey about their initial impressions.
`
`We found users to be overwhelmingly positive about using the zooming web browser. We received such comments as:
`
`“I think it’s different than any other browser, and a lot more interesting to use.”
`
`“It will take some getting used to, but to have the ability to create a tree of where you were is a great advan-
`tage.”
`
`“It’s easier and more friendly than Netscape. It is a little slow and jerky too.”
`
`“I found it a very useful browser, and liked the hierarchical tree structures that are created. There is no need
`to click ‘back’ or ‘forward’ like you do for Mosaic or Netscape.”
`
`“Easy access to previously viewed pages-- excellent way to view pages larger than the screen.”
`
`The survey also asked users to select one or more entries from a list of possible short descriptions of their experience. The results
`
`

`
`are given in Table 1:
`
`SHORT DESCRIPTION
`OF ZOOMING
`WEB BROWSER
`
`# OF PARTICIPANTS
`WHO SELECTED IT
`(out of 14)
`
`Interesting
`
`Would try using it again
`
`Enjoyable
`
`Useful
`
`Exceptional
`
`Frustrating
`
`Confusing
`
`12
`
`11
`
`7
`
`7
`
`4
`
`3
`
`2
`
`Would never use it again
`0
`Table 1: User Study results
`
`When test participants were asked to describe what they most liked about using the zooming web browser, over 70% of the par-
`ticipants said they liked seeing the tree of where they were in the WWW and navigating by zooming. When participants were
`asked to describe what they liked the least, they commonly mentioned the speed of interaction. Users would like to have faster
`browsing tools, as well as the ability to delete a page from the tree structure. Based on this feedback, we have since modified the
`browser. In the newest release, WWW pages are displayed several times faster than during the pilot test and users now have the
`option of deleting any viewable page on the tree.
`
`In the pilot test survey, participants were also asked how they thought the zooming web browser compared with other WWW
`browsers. Of the 14 test participants, a little over half felt qualified to respond. (A handful of the College of Education partici-
`pants had never used another browser and therefore could not make a comparison.) The participants who responded were over-
`whelmingly more positive about the zooming web browser than Netscape or Mosaic. They felt that the browser had a better
`visual layout and was generally easier to use than other browsers. Surprisingly, only one participant felt that Netscape was easier
`and less buggy.
`
`In summary, the results of our pilot test offer positive support for and constructive feedback about the use of zooming in a web
`browser. This feedback continues to inform our browser development efforts. We expect to continue testing with a more diverse
`population of users to better understand the problems and advantages issues associated with zooming.
`
`5. A USER SCENARIO FOR THE FUTURE
`With the technologies we are currently creating, we can foresee a time in the future when the following scenario will come to
`pass:
`
`It is morning. David Brooks enters his classroom. In an hour his fifth grade students will join him, but until then, David
`sits down at his computer, coffee in hand, to scan his favorite web pages. David begins by wandering the local museums’
`home pages. He knows that today he and his students will begin a thematic unit on dinosaurs. Before they arrive, he
`quickly drags various dinosaur images and text from different web pages, and creates a new student page. He decides that
`his page looks more like a wall of graffiti than a planned document. He wishes he had more time to animate the dino-
`saurs, design information lenses, and establish zooming links to other home pages. Suddenly it dawns on him, those
`would be great things for his students to do! He breaths a sigh of relief, sips his coffee, and waits for his students to
`arrive.
`
`Once all 21 students have been welcomed, David explains that thanks to their insistence, they will now turn their energies
`to learning about dinosaurs. The students clap and cheer. Once they settle down, David splits them up into design teams
`
`

`
`of 3, and asks them to move to their computers. On each student team’s screen is an image of David’s dinosaur page.
`David explains how bad this page is, and asks his students to help him redesign it. He asks them to create animations,
`lenses, and links to other pages. The students eagerly work on their projects.
`
`A week later, the student teams present their work to the class. The first group to go presents a page with four simply
`drawn dinosaurs. When a student zooms into the web page, the dinosaurs begin to move about. One of the students
`points out that they found information on the web that described how these dinosaurs moved, so they designed their
`beasts with this in mind. The team zooms in on one dinosaur, a tyrannosaurus rex. As they zoom in, the picture of the
`dinosaur disappears and text information appears. The student team explains that by selecting the highlighted word in the
`text it will take you back to the original WWW page that the text came from. As they explain, the text zooms out, a new
`tree link is formed, and a new page is zoomed in on the screen.
`
`After much applause the next student team presents their work. They zoom in on their WWW page to display one large
`dinosaur. They explain that their project gives you lots of information on just one dinosaur. They begin by dragging var-
`ious lenses from the side of their page. Each lens is shaped like the information it displays. The large tree-shaped lens
`when dragged over the dinosaur shows the types of vegetation the dinosaur eats. The sun-shaped lens shows the kinds of
`climates this dinosaur likes to live in. The team explains that by zooming in with any of the lenses it will take you to a
`WWW page with more information. They demonstrate this and start to uncover a tree of information.
`
`David is impressed with his class’s work. As each team presents, they offer more creative solutions than he thought pos-
`sible. At the end of class he decides to ask the school principal if he can publish his students’ research projects on the
`WWW. Not only does she agree, but she explains that the local museum has been looking for WWW pages created by
`students. She points out that this would be just the thing for their WWW section entitled: “A Kid’s Tree of Knowledge”
`that is being designed to commemorate the fiftieth anniversary of Vannevar Bush’s As We May Think article.
`
`6. CONCLUSION
`The World-Wide Web has become an important and widely used resource. Because of this, it is crucially important to addr