User Interface Tailoring using Semantic Information
Guido Menkhaus
Software Research Lab,
Menkhaus@SoftwareResearch.net
Abstract
Due to the diversity of display capabilities, mobile computing devices have caused a dramatic increase in the development effort of interactive services. User interface tailoring represents a promising concept for coping with this challenge. The position paper proposes a hybrid approach to the generation of adaptive user interfaces based on syntactic and semantic information. Syntactic information is represented by the user interface's size requirements and the semantic information is integrated with a semantic linking language.
1 Introduction
A main obstacle to device-independent content authoring is the target devices' different presentation capacity. The question that needs to be solved is: How to enable content to be tailored to various devices' presentation capacity in order to minimize the mismatch between its presentation and the device's capacity to present it.
A lot of recent work in the area of device-independent authoring has been dedicated to user interface transcoding systems. Transcoding is the process where content is translated or converted from one representation to another. However, the content is generally visualized unaltered on each platform, from a compositional, navigational layout point of view. Assuming a general model that partitions user interfaces into windows and widgets, a window is generally visualized unaltered on each platform, from a compositional point of view, although using different platform specific widgets. We state that it is insufficient splitting a window uniquely on the basis of the size of the widgets into a new set of windows. Without additional semantic information this results in a set of windows that can only be navigated in a linear, sequential way. To access the last widget of the original window, each window of the linear set has to be traversed. This process is often called fragmentation [1] or pagination [2].
The challenge is to remodel the widgets of a window into a new composition of windows with a reasonable flow of transitions between them (c.f. Figure 1) by integrating size and semantic information between widgets. Semantic information may be expressed with relations such as " A depends on B ", " A contains B " etc..
2 Principle
User interfaces
should be tailored to accommodate the different presentation capacity of a
user's device. Device independent content authoring needs content to be
tailored according to
- the presentation capacity characteristic of the device (e.g.,
device's screen size) and
- 2. the semantic relation between widgets (c.f. Figure 2).
However, no
markup-based user interface description language allows the definition of
semantic relation between widgets.
3 Concepts
The concepts presented in this section help answering the following aspects of user interface tailoring:
- Mechanism to tailor a user interface.
- Integration of semantic information into a markup-language based user interface description.
3.1 Clustering Widgets
Instead of transforming a window into a linear, linked list of window, where each window is tailored to the size characteristics of the target device, the clustering mechanism applies a transformation that results in a hierarchical structure. The clustering process is guided by
- Device presentation capacity (e.g., device's screen size) and
- Semantic information.
The clustering process groups widgets as long as the device display size allows it and the semantic information does not oppose it into windows. Large-sized widgets may be splitted into a linear list of windows. Additional semantic information enables the clustering process generating a user interface having a hierarchical navigational layout structure. The hierarchical structure allows better navigation and gives the user a better overview than the linear, linked list of windows.
3.2 Semantic Links
Semantic information about the relation between widgets is inserted into a user interface description language with what might be called the XML Semantic Linking Language (XSLink), which needs to be defined. XSLink works similarly to XLink [3]. XLink allows elements to be inserted into XML documents in order to create and describe links between resources. XSLink would allow describing relations between elements in XML documents. The description results in a semantic net, on which the clustering process is based. The semantic net guides the clustering process: Depending on the semantic net, the clustering process results in a different hierarchical structural and navigational layout.
4 Summary of what would be presented if invited
- A hybrid approach to the generation of adaptive UIs based on a linking strategy of hierarchies of graphs.
- Results of the hybrid approach: User interfaces under different device presentation capacities on mobile telephones.
References
K.H. Britton, R.Case, A.Citron, R. Floyed, Y. Li, C. Seekamp, B. Topol, and K. Tracey. Transcoding. Extending e-business to new environments. IBM Systems Journal, 40(1):153-178, 2001. http://researchweb.watson.ibm.com/journal/sj/401/britton.html.
Microsoft. Mobile Internet Toolkit - QuickStart Tutorial, 2002. http://www.gotdotnet.com/MobileQuickStart.
W3C. XLink 1.0. In MSteve DeRose, Eve Maler, and David Orchard, editors, W3C Technical Reports and Publications. W3C, 2001. http://www.w3.org/TR/xlink/.
File translated from
On