User Interface Design Options

4.2.3Mapping Issues: Attribute and Display Interactions

• 
  too many dimensions to be displayed -- reduction of the attribute set size to a reasonable number
  
• 
  incidental or misleading information mapping -- choosing the right data and display types
  
• 
  implied or lost data resolution-- choosing the right discrimination
  
• 
  impossible discriminations -- paying attention to task/perceptual capabilities
  
• 
  ambiguous encoding -- paying attention to unintended encoding
  
• 
  data loss in the display -- paying attention to the limits of the hardware
  

When the number of attributes to be displayed is high, the fixed number of dimensions available on the display can present a problem. Additionally, even if all the attributes can be displayed, correlation between various attributes may result in an implied causality in the display. Data analysis techniques should be applied to remove unneeded attributes. Task analysis can indicate the important attributes to be displayed.

We begin with a statement that is implied but not explicitly stated above about the data type and discriminability mapping rules. Mapping nominal data to an ordinal or interval dimension will imply characteristics of the objects that are not true. For example, if the nationality of a publication is mapped to size, there is some implication that one country's publications are “smaller” than another's. Similarly, if age is mapped to hue value of color, there is an implication that there is no ordering to age.

Mapping data requiring fine discrimination to a dimension that does not allow fine discrimination results in a data loss. In contrast, mapping low discrimination data to a high discrimination dimension can result in false distinctions. A screen has some limits both in size and resolution. Holding in abeyance for the moment the fact that the screen may be a viewport onto a larger space, we find a limit of about 1,000,000 pixels on modern screens. While the resolution of these individual pixels on a true color display is theoretically 2²⁴, (16million) colors, the reality is that most humans would have trouble discriminating more than 256 colors. With improvements in display size and resolution, human visual acuity will set the upper bound of displayed data density.

Having suggested some simple measure of resolution, it is important to note that perceptual sensitivity varies widely based on the specific task. Identification, comparison, and recall all have differing levels of sensitivity. The human factors and cognitive science literature record a variety of different studies on visual sensitivity. Among the factors that have been studied are:

• 
  The threshold of “just noticeable differences” for display attributes such as position, length, size, etc.
  
• 
  The variation in “just noticeable differences” dependent upon the intensity of signal.
  
• 
  The ability to discriminate absolute and relative values. (It is easier to discriminate values when they are side by side than when we need to remember and make a judgment about the values.)
  

There are a number of situations in which displayed information may carry unintended information. Consider for example a hierarchical display of the files in a directory. The files must be listed in some order. It is possible that the order in which they are presented may imply some meaning that is not actually present. Ambiguity may also be introduced when the complexity of the encoding is high, for example, if the meaning of one display attribute changes based on the value of another attribute; e.g., for active documents, the hue represents X while for inactive documents, the hue represents Y. Interaction between encoded attributes may also lead to ambiguity; e.g., shape and size represent two attributes: when size is small, shapes become indistinguishable or disappear.

If a document attribute causes data to be mapped into exactly the same location as previously stored data, the existing data may be lost. An example would be an XY graph plot of data points which have the same value for X and for Y. Ideally, some indicator will be in force to indicate the existence of multiple, superimposed objects. Nonetheless, occlusion and object overlap present potentially serious problems of data loss. In a similar fashion, data that is plotted outside the clipping boundary of the screen is lost.

4.2.4Interaction

4.2.4.1Global manipulation

When objects are mapped into a screen or window space, it is possible that the full set will not fit in the given display space without crippling the value of the display. Given the limited space for display, graphical transformations can be used to show various objects. These manipulations help in the following ways: to solve the clipping problem by allowing panning (translation); to solve the occlusion problem by changing the view point (rotation); and to solve the resolution problems by zooming (scaling). A few of the more common techniques are:

• 
  Pan and Zoom: Given an unlimited virtual space (2 dimensional), this technique allows the user to pan across the space and zoom in on a given point. The manipulation changes the view of the display space.
  
• 
  Flyby: When the screen provides a viewport onto a 3D space from a given point of view, translation along with rotation and scaling of the space can be used to change the user’s view of the space. Done effectively, this provides a virtual reality kind of simulation of a flyby of a virtual space.
  
• 
  Distortion plane: Given a space that shows all objects at once, it may not be possible to see any particular object well enough. To solve this problem, we use a non-linear scaling of the plane. A summary and unified theory of distortion-orientation of a presentation can be found in Leung and Apperley (1994). Examples of distortion plane techniques include:
  
• 
  Perspective wall: uses flat zoom
  
• 
  Fish eye: applies parabola zoom function
  
• 
  Multi-points distortion: The distortion does not have to be a single point at center. It can be many points at any place on the display space.
  
• 
  Zoom with more attributes: When zooming, objects spread apart in space; that space can be used to display more attributes, i.e. increasing detail of objects. It is also called semantic zoom.
  
• 
  Change mode display. This technique changes the presentation. Most simply the same objects and perspectives are presented, but using a different encoding scheme from the first one, using the same screen space.
  

4.2.4.2Local manipulation

Several different kinds of local manipulation are possible as well. Generally, they are used to make views on a data set clearer. They include the following:

• 
  Object repositioning: A given object or set of objects in a display may be moved. This process may reveal other objects or clarify other relationships. Moving depends on the encoding scheme. Free space moving is allowed in non-XY encoding. Systems using XY encoding may enforce a relocation scheme. It is also possible that selected objects in a view may have the impact of causing other objects to react, making information about these objects available as well.
  
• 
  Object re-rendering by scale manipulation: The scale used on a given dimension may be changed. For example, the x dimension may be shifted from a linear to a logarithmic scale, changing the location at which objects appear. Similarly, the mapping of the color attribute of a document may be shifted by a technique such as histogram equalization in order to better display the difference among clustered objects.
  
• 
  Expanding and collapsing: It is commonly used in hierarchical structures; the children may be shown or not, depending on interaction at the parent level.
  

4.2.5Multiple-Views

Multiple views of the data set can be used to reduce ambiguity. The views can be differentiated in several ways. They can be viewed simultaneously or sequentially. They can be different presentations of the same data set or consonant presentations of different data sets. In this latter case, the data sets can be disjoint, overlapping, or subsets. Of course, none of these are either/or possibilities and several of the conditions can be in force at a given point in time. A few of the more interesting multiple views are:

• 
  Multiple Window Views
  
• 
  Global View: a window provides an overview of some space or data set and an indication of what portion of the overall view is being displayed in a second window.
  
• 
  Object/Attribute View: a window provides details of the attributes of a selected object in the second display.
  
• 
  Single Window Views
  
• 
  Data Delta View: a window shows a different view of the same scope of space, and attempts either by temporal proximity or some other animation method to show the displacement of the data objects.
  
• 
  Drill Down View: a window shows a set of objects and has rules for expanding the objects. When a selected object can be expanded, the results of the expansion either replace or are added to the original window.
  

4.2.6Sample Interaction Techniques

There are virtually unlimited combinations of encoding scheme and interaction. Many of the encoding schemes and interactions are defined by the developer. Some popular interaction techniques are:

• 
  Dynamic Queries: Dynamic queries, developed by Shneiderman (1993), is an interactive display with controllers for direct-manipulation of queries and results. Controllers are created which are bound to a range of values of interest corresponding to an attribute. The presented data changes dynamically as the controller is manipulated in a bounded range.
  
• 
  Mural: Mural is a scheme that provides an overview presentation to be fit on the screen (Jerding & Stasko, 1995). The Mural view is a miniature of a larger content that cannot be viewed without losing detail or is not readable in a single display space. The display space is condensed; therefore, a single dot or Mural view may present multiple data points from the original display. A secondary encoding, such as color dimension of pixels, may be used to present additional data.
  
• 
  Magic Lenses: A Magic Lens uses the concept of spatial sensitivity. It mimics a magnifying lens. Magic lenses are areas which are superimposed on top of another presentation. Many functions can be applied to lenses. What is shown on the lens is a function of the lens's position (Stone, Fishkin, & Bier, 1999).
  
• 
  Pad and Pad++: Pad uses zooming and panning as a main interaction (Perlin & Fox, 1993;Bederson & Hollon, 1994). The data first appear at a certain magnification factor. Zooming in shows objects in a bigger size, with more detail, or with different detail (semantic zooming). An object in Pad space is 3D, X-Y coordinate and depth. At a certain zoom factor, objects that have a certain depth will appear.
  
• 
  Furnas' Fisheye: Furnas (1981) provides a view combination showing both overview and detail; using zooming, an overview structure is not visible when viewing detail. Using multiple views, it is difficult to relate the information in both views. The fisheye provides a combination of overview and detail views in a single view. The objects of interest will be visible and dynamically change according to the focus point. By using the distance metaphor, the focal object will appear closer than other objects. It is a basis concept of distortion presentation of space by “degree of interest” and it also called “Detail+Context” or “Focus+Context.” Formal description and framework of “Focus+Context” is in Björk, Holmquist, and Redström (1999).
  

Download 324.2 Kb.

Share with your friends: