over pie menu to display thumbnails of all windows that lie beneath that selected pixel.
However, the techniques only considered the windows close to the mouse cursor, and when the pie menu displays multi items, it becomes difficult to display thumbnails of each window at a recognizable size.
way (automatic way), by contrast, windows are organized to its most likely group based on upon evidence group creation and manipulation, such as semantic data about the windows (window title, window content), activation history. The big problem of this way is that the system can incorrectly predict the users’ intention and users can fail to understand the system prediction (e.g. WindowScape Tashman(2006)). In the following paragraphs, we first describe related works of explicit window grouping techniques, then reviewing current knowledge of implicit window grouping techniques.
2.3.1 Explicit Window Grouping Techniques
For explicit window grouping technique, users can define groups in any way, and it requires to plan in advance the number of groups and associate each window to a group.
There are many techniques that have been proposed.
2.3.1.1 Virtual Desktop Managers
Virtual desktop managers (VDMs) have been one of the most popular solutions to explicitly define groups but at the cost of a strict separation between them, making it difficult to switch between windows from multiple groups. Meanwhile VDMs have limi- tations in the flexibility of grouping mechanisms they provide, and the means offered for finding particular windows. One of the earliest designs exploring a VDMs was Smalltalk Project Views Goldberg & Robson(1983). Rooms D. Austin Henderson & Card (1986) is probably the most famous of these kinds of systems. A number of these systems are currently available, and are described in XDESK. None of these systems have been eval- uated in a stringent way and no detailed study demonstrate how easy to learn they are, or how well they integrate into real-world settings.
2.3.1.2 GroupBar
In addition to VDMs, a number of novel and more flexible solutions have been pro- posed, such as GroupBar Smith et al. (2003), Scalable Fabric Robertson et al. (2004) and SCWM Badros et al. (2000). GroupBar (Figure 2.13) used the same minimized window representation used by Windows Taskbar to task creation and maintenance by allowing users to create or remove a task by using a single dragging gesture and allowed users to simultaneously display any subset of windows, even if they should be assigned to
2.3 Group Switching Techniques
Figure 2.13: GroupBar is a prototype for demonstrating the use of window-grouping features in an Windows XP TaskBar-like interface.
different tasks. The design did not effectively leverage human spatial and visual recogni- tion memory. GroupBar made limited use of spatial memory by allowing users to create multiple bars, limitations stem from the bar design, which is linear, list-based, and did not expose much virtual space in which to place tasks.
2.3.1.3 Elastic Windows
Kandogan et al. proposedElastic Windows that was designed to provide an alterna- tive to window management strategies for efficient personal role management based on hierarchical window organization, multi-window operations and space-filling tiled layout Kandogan & Shneiderman (1996). In Elastic Windows, window groups could be created by opening a container window and dragging and dropping the selected items inside this window. The authors compared Elastic Windows to an independent overlapping window
Figure 2.14: Scalable Fabric showing the representation of three tasks as clusters of win- dows, and a single window being dragged from the focus area into the periphery.
manager in terms of user performance time on task environment, switching, and four task execution types. They found almost in all cases that Elastic Windows resulted in faster task completion and task switching times.
2.3.1.4 Scalable Fabric
Scalable Fabric (Figure 2.14) makes use of the periphery of the display for spatial layout of tasks, in addition to leveraging users’ efficient visual recognition memory for images, allocating screen real estate in accordance with a user’s attention, using a focus- plus-context display, allowing users to group collections of windows that are used together and to leave windows and clusters of windows open and visible at all times via a process of scaling down and moving the windows and clusters to the periphery.
However users need to leverage the size of the focus area and the periphery in order to operate (the focus area) and recognize (the periphery) the windows.
2.3.1.5 SCWM
The key feature ofSCWM is the user’s ability to interact with groups of windows by setting constraints on the windows, the constraints aim for windows layout, and lack of control over windows stacking. Users create these constraints by using a relationship panel that graphically represents the types of constraints that can be set. Using the panel hides the details of the constraint system. An example constraint is adjacency, where some
2.3 Group Switching Techniques
window A is constrained to be adjacent to some other window B. Then, whenever a user repositions either A or B, its partner is simultaneously repositioned without additional user interaction. Unfortunately, there is not enough information about how users currently use SCWM and where and how it is adequate.
A disadvantage faced by these systems is that requiring windows to be in a single group forces users to decide ahead of time where a new window belongs. These systems do not allow for windows to reside in multiple groups at the same time. Likewise, such a single-group approach is inflexible, in that it neglects the possibility that some windows might naturally be useful in several different groups.
2.3.2 Implicit Window Grouping Techniques
Compared to explicit window grouping techniques, implicit grouping techniques use different methods, such as machine learning, to automatically define groups. However the system can incorrectly infer groups and create groups that may not correspond to users expectations. There are also many techniques that have been proposed.
2.3.2.1 WindowScape
WindowScape Tashman (2006) (Figure2.15) is an example of implicit grouping tech- nique which automatically creates groups by taking photograph-like snapshots each time a window is expanded or miniaturized, using the timeline of desktop states shown as a series of photograph-like snapshots. However, when a user wants to resume a group, it may no longer be visible or the user may have to explicitly define favorite snapshots. Each snapshot group can include many windows and two close groups may include the same windows (windows have the same state (position, size)), this can lead to the snapshots that look like very similar, thereby it may cause difficulties to distinguish between the two groups. Similarly, WindowScape does not appear to have been evaluated.
2.3.2.2 Stack leafing
Stack leafing Faure et al. (2009) technique is another example of implicit grouping techniques, which is based on widget that combines generalized scrolling and crossing to control the stacking order of layers of non-overlapping windows. This technique has the advantage of minimizing mouse navigation and preserving the size and position of
Figure 2.15: WindowScape represents windows as thumbnails and uses the timeline of desktop states shown as a series of photograph-like snapshots.
windows while still providing access to all of their content. However, this technique does not preserve the ordering by frequency and strict non-overlapping requirement also limits its use.
2.3.2.3 SWISH
SWISH Oliveret al.(2006) uses semantic features (windows associated with the same task share common words in their content, and, in particular, in their window titles) and temporal features (windows associated with the same task are accessed in temporal proximity to each other) to automatically divide windows into different groups. Their evaluations suggested 70% accuracy rates in assigning windows to task groups.