Link

Murielle Florins Université catholique de Louvain, Louvain-la-Neuve, Belgium
Jean Vanderdonckt Université catholique de Louvain, Louvain-la-Neuve, Belgium

International Conference on Intelligent User Interfaces
Proceedings of the 9th international conference on Intelligent user interface
Funchal, Madeira, Portugal
SESSION: Multi-platform interfaces table of contents
Pages: 140 - 147
Year of Publication: 2004
ISBN:1-58113-815-6

Abstract:
This paper introduces and describes the notion of graceful degradation as a method for supporting the design of user interfaces for multiplatform systems when the capabilities of each platform are very different. The approach is based on a set of transformational rules applied to a single user interface designed for the less constraint platform. A major concern of the graceful degradation approach is to guarantee a maximal continuity between the platform specific versions of the user interface. In order to guarantee the continuity property, a priority ordering between rules is proposed. That ordering permits to apply first the rules with a minimal impact on the multiplatform system continuity.

My Discussion:
This paper is strong in its overview of the thought and current work being done on ways to create a User Interface that can be used on multiple platforms, especially when the platforms differ strongly in capabilities, like workstations versus mobile phones. After the work in the early and mid-nineties where everything was about lowest-common denominator late-binding toolkits like Visix Galaxy and AWT, this paper also (briefly) lists the more recent approaches in cross-platform UIs as afforded by the XML / CSS revolution or XML transformations through XSLT, and UIs generated from high-level task descriptions like UIML. The paper then proceeds to explain its own method of creating UIs for platforms with different capabilites: examine a UI on three, increasingly abstract levels, being the 1) the Concrete UI which is an abstract view on the UI in terms of layout and what capabilities widgets in the lay-out have, 2) the Abstract UI, which is a view on the UI on the level what 'presentation units', here defined as a "presentation environment (e.g. a window or a panel) that supports the execution of a set of logically connected tasks", and 3) the Tasks and Concepts level. It then defines transformation rules for these levels, like "replace a widget used on the high-level platform for a simpler equivalent widget on the phone" for level 1 and "split a big window into smaller presentation units when going to smaller screens" for level 2 and "don't offer the user to perform a task that can't be done at all on smaller screens" for level 3. After describing this collection of rules, they are then prioritzed so as to create as little change in the UI when transforming it for less capable platforms while still ending up with a usable UI for the target. Of course, the rules thus end up being prioritized by level, rules on the lay-out and widget level to be applied first and rules like dropping whole tasks last. And that's, alas, where the paper stops. No discussion on how hard it is to automate these rules, on tools being worked on, if these rules even made sense when applying them to a UI process by a designer by hand as a sort of heuristic. Great collection of refernces, though.