The Eyes Have It:
A Task by Data Type Taxonomy for Information Visualizations
Department of Computer Science,
Copyright By PowCoder代写 加微信 powcoder
Human-Computer Interaction Laboratory, and Institute for Systems Research University of Maryland
College Park, Maryland 20742 USA
ben @cs.umd.edu
A usefulstartingpointfordesigningadvancedgraphical user interjaces is the Visual lnformation-Seeking Mantra: overview first, zoom and filter, then details on demand. But this is only a starting point in trying to understand the rich and varied set of information visualizations that have beenproposedinrecentyears. Thispaperoffersataskby data type taxonomy with seven data types (one-, two-, three-dimensional datu, temporal and multi-dimensional data, and tree and network data) and seven tasks (overview, Zoom, filter, details-on-demand, relate, history, and extracts).
Everything points to the conclusion that the phrase ‘the language of art’ is more than a loose metaphor, that even to describe the visible world in images we need a developed system of schemata.
keys), are being pushed aside by newer notions of information gathering, seeking, or visualization and data mining, warehousing, or filtering. While distinctions are subtle, the common goals reach from finding a narrow set of items in a large collection that satisfy a well-understood information need (known-item search) to developing an understanding of unexpected patterns within the collection (browse) (Marchionini, 1995).
Exploring information collections becomes increasingly difficult as the volume grows. A page of information is easy to explore, but when the information becomes the size of a book, or library, or even larger, it may be difficult to locate known items or to browse to gain an overview,
Designers are just discovering how to use the rapid and high resolution color displays to present large amounts of information in orderly and user-controlledways. Perceptual psychologists, statisticians, and graphic designers (Berlin, 1983; Cleveland, 1993; Tufte, 1983, 1990) offer valuable guidance about presenting static information, but the opportunity for dynamic displays takes user interface designers well beyond current wisdom.
2. Visual Information Seeking Mantra
The success of direct-manipulation interfaces is indicative of the power of using computers in a more visual or graphic manner. A picture is often cited to be worth a thousand words and, for some (but not all) tasks, it is clear that a visual presentation-such as a map or photograph-is dramatically easier to use than is a textual description or a spoken report. As computer speed and display resolution increase, information visualization and graphical interfaces are likely to have an expanding role. If a map of the United States is displayed, then it should be possible to point rapidly at one of 1000 cities to get tourist information. Of course, a foreigner who knows a city’s name (for example, ), but not its location, may do better with a scrolling alphabetical list.
E. H. Gombrich Art and 1. Introduction
1959 (p. 7 6 )
Information exploration should be a joyous experience, but many commentators talk of information overload and anxiety (Wurman, 1989). However, there is promising evidence that the next generation of digital libraries for structured databases, textual documents, and multimedia will enable convenient exploration of growing information spaces by a wider range of users. Visual language researchers and user-interface designers are inventing powerful information visualization methods, while offering smoother integration of technology with task.
The terminology swirl in this domain is especially colorful. The older terms of information retrieval (often applied to bibliographic and textual document systems) and database management (often applied to more structured relational database systems with orderly attributes and sort
0-8186-7469-5/96 $05.00 0 1996 IEEE
Authorized licensed use limited to: University of Glasgow. Downloaded on May 25,2021 at 13:18:37 UTC from IEEE Xplore. Restrictions apply.
Visual displays become even more attractive to provide orientation or context, to enable selection of regions, and to provide dynamic feedback for identifying changes (for example, a weather map). Scientific visualization has the power to make ,atomic, cosmic, and common three- dimensional phenomena (such as heat conduction in engines, airflow aver wings, or ozone holes) visible and comprehensible. 14bstract information visualization has the power to reveal patterns, clusters, gaps, or outliers in statistical data, stock-market trades, computer directories, or document collections.
Overall, the bandwidth of information presentation is potentially higher in the visual domain than for media reaching any of the other senses. Humans have remarkable perceptual abilities,that are greatly under-utilized in current designs. Users can scan, recognize, and recall images rapidly, and can detect changes in size, color, shape, movement, or texture. They can point to a single pixel, even in a megapixel display, and can drag one object to another to perforrn an action. User interfaces have been largely text-oriented, so as visual approaches are explored, appealing new opportunities are emerging.
There are many visual design guidelines but the basic principle might be: summarized as the Visual Information Seeking Mantra:
Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand Overview first, zoom and filter, then details-on-demand
Each line represents one project in which I found m:yself rediscoveriing this principle and therefore wrote it down it as a reminder. It proved to be only a starting point in trying to characterize the multiple information- visualization innovations occurring at university, government,and industry research labs.
3. Task by Data Type Taxonomy
To sort out the prototypes and guide researchers to new opportunities, I propose a type by task taxonomy (TTT) of information viisualizations. I assume that users are viewing collections of items, where items have multiple attributes. In all seven data types (1-, 2-, 3-dimensional data, temporal and multi-dimensional data, and tree and network data) the items have attributes and a basic search task is to select all items that satisfy values of a set of
attributes. An example task would be finding all divisions in an organization structure that have a budget greater than $500,000.
The data types are on the left side of the TTT characterize the task-domiain information objects and are organized by the problems users are trying to solve. For example, in two-dimensilonal information such as maps, users are trying to grasp adjacency or navigate paths, whereas in tree-structureclinformation users are trying to understand parent/child/sibling relationships. The tasks across the top of the TTT are task-domain information actions that users wish to perform.
The seven tasks are at a high level of abstraction. More tasks and refinements of these tasks would be natural next steps in expanding this table. The seven tasks are:
Overview: Gain an overview of the entire collection. Zoom : Zoom in on items of interest
Filter: filter out uninteresting items. Details-on-demand:Select an item or group and get
details when needed.
Relate: View relations hips among items.
History: Keep a history of actions to support undo,
replay, and progressive refinement.
Extract: Allow extraction of sub-collections and of the
query parameters.
Further discussion of the tasks follows the descriptions of the seven data types:
1-dimensional: linear data types include textual documents, program source code, and alphabetical lists of names which are all organized in a sequential manner. Each item in the collection is a line of text containing a string of characters. Additional line attributes might be the date of last update or authior name. Interface design issues include what fonts, color, size to use and what overview, scrolling, or selection methods can be used. User problems might be to find the number of items, see items having certain attributes (show only lines of a document that are section titles, lines of a program that were changed from the previous version, or people in a list who are older than 21 years), or see an item with all its attributes.
Examples: An early approach to dealing with large 1- dimensional data sets was the bifocal display which provided detailed information in the focus area and less information in the surrounding context area (Spence and Apperley, 1982).In their example, the selected issue of a scientific journal had detiails about each article, the older and newer issues of the journal were to the left and right on the bookshelf with decreasing space. Another effort to visualize 1-dimensionaldata showed the attribute values of each thousands of item in a fixed-sized space using a
Authorized licensed use limited to: University of Glasgow. Downloaded on May 25,2021 at 13:18:37 UTC from IEEE Xplore. Restrictions apply.
scrollbar-like display called value bars (Chimera, 1992). Even greater compressions were accomplished in compact displays of tens of thousands of lines of program source code (SeeSoft, Eick et al., 1992) or textual documents (Document Lens, Robertson and Mackinlay, 1993; Information mural, Jerding and Stasko, 1995).
2-dimensional: planar or map data include geographic maps, floorplans, or newspaper layouts. Each item in the collection covers some part of the total area and may be rectangular or not. Each item has task-domain attributes such as name, owner, value, etc. and interface- domain features such as size, color, opacity, etc. While many systems adopt a multiple layer approach to dealing with map data, each layer is 2-dimensional. User problems are to find adjacent items, containment of one item by another, paths between items, and the basic tasks of counting, filtering, and details-on-demand.
Examples: Geographic Information Systems are a large research and commercial domain (Laurini and Thompson, 1992; Egenhofer and Richards, 1993) with numerous systems available. Information visualization researchers have used spatial displays of document collections (Korfhage, 1991; Hemmje et al., 1993; Wise et al., 1995)
organized proximally by term co-occurrences.
3-dimensional: real-world objects such as molecules, the human body, and buildings have items with volume and some potentially complex relationship with other items. Computer-assisted design systems for architects, solid modelers, and mechanical engineers are built to handle complex 3-dimensional relationships. Users’ tasks deal with adjacency plus above/below and inside/outside relationships, as well as the basic tasks. In 3-dimensional applications users must cope with understanding their position and orientation when viewing the objects, plus the serious problems of occlusion. Solutions to some of these problems are proposed in many prototypes with techniques such as overviews, landmarks, perspective, stereo display, transparency, and color coding.
Examples: Three-dimensional computer graphics and computer-assisted design are large topics, but information visualization efforts in three dimensions are still novel. Navigating high resolution images of the human body is the challenge in the National Library of Medicine’s Visible Human project (North et al., 1996). Some applications have attempted to present 3-dimensional versions of trees (Robertson et al., 1993), networks (Fairchild et al., 1988), or elaborate desktops (Card et al., 1996).
Temporal: time lines are widely used and vital enough for medical records, project management, or
historical presentations to create a data type that is separate from 1-dimensional data. The distinction in temporal data is that items have a start and finish time and that items may overlap. Frequent tasks include finding all events before, after, or during some time period or moment, plus the basic tasks.
Examples: Many project management tools exist, but novel visualizations of time include the perspective wall (Robertson et al., 1993) and LifeLines (Plaisant et al., 1996). LifeLines shows a youth history keyed to the needs of the Maryland Department of Juvenile Justice, but is intended to present medical patient histories as a compact overview with selectable items to get details-on-demand. Temporal data visualizations appear in systems for editing video data or composing animations such as Macromedia Director.
Multi-dimensional: most relational and statistical databases are conveniently manipulated as multi- dimensional data in which items with n attributes become points in a n-dimensional space. The interface representation can be 2-dimensional scattergrams with each additional dimension controlled by a slider (Ahlberg and Shneiderman, 1994). Buttons can used for attribute values when the cardinality is small, say less than ten. Tasks include finding patterns, clusters, correlations among pairs of variables, gaps, and outliers. Multi-dimensional data can be represented by a 3-dimensional scattergram but disorientation (especially if the users point of view is inside the cluster of points) and occlusion (especially if close points are represented as being larger) can be problems. The technique of parallel coordinates is a clever innovation which makes some tasks easier, but takes practice for users to comprehend (Inselberg, 1985).
Examples: The early HomeFinder developed dynamic queries and sliders for user-controlled visualization of multi-dimensional data (Williamson and Shneiderman, 1992). The successor FilmFinder refined the techniques (Ahlberg and Shneiderman, 1994) for starfield displays (zoomable, color coded, user-controlled scattergrams), and laid the basis for the commercial product Spotfire (Ahlberg and Wistrand, 1995). Extrapolations include the Aggregate Manipulator (Goldstein and Roth, 1994), movable filters (Fishkin and Stone, 1995), and Selective Dynamic Manipulation (Chuah et al., 1995). Related works include VisDB for multidimensional database visualization (Keim and Kreigal, 1994), the spreadsheet-like Table Lens (Rao and Card, 1994) and the multiple linked histograms in the Influence Explorer (Tweedie et al., 1996).
Tree: hierarchies or tree structures are collections of items with each item having a link to one parent item (except the root). Items and the links between parent and
Authorized licensed use limited to: University of Glasgow. Downloaded on May 25,2021 at 13:18:37 UTC from IEEE Xplore. Restrictions apply.
child can have muKtiple attributes. The basic tasks can be applied to items and links, and tasks related to structural properties become interesting, for example, how many levels in the tree? or how many children does an item have? While it is possible to have similar items at leaves and internal nodes, it is also common to find different items at each level in a tree. Fixed level trees with all leaves equidistant from the root and fixed fanout trees with the same number of children for every parent are easier to deal with. High fanout (broad) and small fanout (deep) trees are important special cases. Interface representations of trees can use an outline style of indented labels used in tables of contents (Chimera and Shneiderman, 1993), a node and link diagram, or a treemap, in which child items are rectangles nested inside parent rectangles.
Examples: Tree-structured data has long been displayed with indented outlines (Egan et al., 1989) or with connecting lines as in many computer-directory file managers. Attempts to show large tree structures as node anld link diagrams in compact forms include the 3- dimensional cone and cam trees (Robertson et al., 1993; Carriere and Kazman, 1995), dynamic pruning in the TreeBrowser (Kumar et al., 1995), and the appealingly animated hyperbolic trees (Lamping et al., 1995). A novel space-filling mosaic approach shows an arbitrary sized tree in a fixed rectangular space (Shneiderman, 1992; Johnson and Shneiderman, 1991). The treemap approach was successfully applied to computer directories, sales data, business decision-making (Asahi et al., 1995), and web browsing (Mitchell et al., 1995; Mukherjea et al., 1995), but users take 10-;!0 minutes to accommodate to complex treemaps.
Network: sometimes relationships among items cannot be convenie:ntly captured with a tree structure and it is useful to have items linked to an arbitrary number of other items. While many special cases of networks exist (acyclic, lattices, rooted vs. un-rooted, directed vs. undirected) it seemis convenient to consider them all as one data type. In addhion to the basic tasks applied to items and links, network users often want to know about shortest or least isostly paths connecting two items or trawersing the entire network. Interface representations include a node and link diagram, and a square matrix of the items with the value of a link attribute in the row and column representing a link.
Examples: Network visualization is an old but still imperfect art because of the complexity of relationships and user tasks. Commercial packages can handle small networks or simple strategies such as Netmap’s layout of nodes on a circle with links criss-crossing the central area. An ambitious 3-dimensional approach was an impressive early accomplishment (Fairchild et al., 1988), and new
interest in this topic has been spawned by attempts to visualize the World Wide Web (Andrews, 1995;Hendley et al., 1995).
These seven data types reflect are an abstraction of the reality. There are many variations on these themes (2 1/2 or 4-dimensional data, multitrees,…) and many prototypes use combinations of these data types. This taxonomy is useful only if it facilitates discussion and leads to useful discoveries. Some idea of missed opportunities emerges in looking at the tasks and data types in depth:
Overview: Gain an overview of the entire collection. Overview strategies include zoomed out views of each data type to see the entire collection plus an adjoining detail view. The overview contam a movable field-of-view box to control the contents of the detail view, allowing zoom factors of 3 to 30. Replication of this strategy with intermediate views enables users to reach larger zoom factors. Another popular approach is the fisheye strategy (Furnas, 1986) which haa been applied most commonly for network browsing (Sarlcar and Brown, 1994; Bartram et al., 1995). The fisheye distortion magnifies one or more areas of the display, but zoom factors in prototypes are limited to about 5. Although query language facilities made it difficult to gain an overview of a collection, information visualization interfaces support some overview strategy, or should. Adequate overview strategies are a useful criteria to look for. Along with an overview plus detail (also called context plus focus) view there is a need for navigation tools to pan or scroll through the collection.
Zoom: Zoom in on items of interest. Users typically have an interest in some portion of a collection, and they need tools to enable them to control the zoom focus and the zoom factor. Smooth zooming helps users preserve their sense of position and context. Zooming could be on one dimension at a time by moving the zoombar controls or by adjusting the size of the field-of -view box. A very satisfying way to zoom in is by pointing to a location and issuing a zooming command, usually by clicking on a mouse button for as long a
程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com