Automatically generating personalized user interfaces with Supple
Download 5.78 Mb. View original pdf
|
| 7. Evaluation In this section we examine Supple’s technical capabilities and limitations. 7.1. Versatility We demonstrate Supple’s versatility by exhibiting the range of different types of interfaces it has generated. Earlier in this paper, we presented an interactive map-based interface (Fig. 2), a fully functional email client (Fig. 3), an interface to Amazon Web Services (Fig. 4), an FTP Client (Fig. 5), an interface for controlling lighting, ventilation and audiovisual equipment in a classroom (Fig. 14), and two different print dialog windows (Figs. 10 and 15). In this section, Fig. provides an example of Supple’s customization capabilities on a dialog box for font formatting. Fig. 18 illustrates a range of supported devices the interface for controlling classroom equipment was rendered for such diverse platforms as a touch panel, an HTML browser, a PDA, a desktop computer and a WAP cellphone. Fig. 19 shows a user interface for controlling a stereo rendered on a PDA and on a desktop computer. Fig. 23 shows a Supple reimplementation of Microsoft’s Ribbon interface for Word 2007. Finally, Fig. 33 in the next section, shows a font formatting dialog generated for users with different motor abilities. These examples demonstrate a range of different types of interfaces device control (classroom and stereo, dialog boxes (font formatting, media-based (map, and data-oriented applications (email and the Amazon client). Additionally, compared to previous rule-based approaches, optimization robustly and flexibly handles tradeoffs and interactions between choices indifferent parts of the interface. For example, a rule-based system will likely fail to exploit an increase in screen size (or decrease in interface complexity) by using more convenient but larger widgets. In contrast, Supple ’s search algorithm always selects an interface that is optimal (with respect to the cost function) fora given interface and device specification. Fig. 20 illustrates how Supple robustly degrades the quality of the generated user interfaces as it is presented with devices with progressively narrower screens. 930 K.Z. Gajos et al. / Artificial Intelligence 174 (2010) 910–950 Download 5.78 Mb. Share with your friends:
|