Information Assembly for Automatic Content Adaptation

Information Assembly for Adaptive Display Andrew LAMPERT CSIRO ICT Centre Locked Bag 17 North Ryde, Australia, 1670 Andrew.Lampert@csiro.au Cécile PARIS CSIRO ICT Centre Locked Bag 17 North Ryde, Australia, 1670 Cecile.Paris@csiro.au Abstract In a world where information is increasingly delivered to users via different devices with dramatically different constraints and capabilities, it is becoming crucial to consider how the presentation of information must be adapted to suit specific devices and user contexts. To avoid confusing and disorienting the users as they switch between devices, the content and structure of information should be kept constant while its presentation must be optimised for each device to ensure usability. In this paper, we distinguish between two types of decisions that must be made during the presentation planning stage of an information delivery system: local decisions, which are based only on the content or features of the node itself, and global decisions, which are based on the entire structure of the discourse tree. We present a generic algorithm for making global decisions, driven by the discourse tree structure and contextual characteristics. 1 Introduction In a world where information is increasingly delivered to users via different devices with dramatically different constraints and capabilities, it is becoming crucial to consider how the presentation of information must be adapted to suit specific devices and user contexts. To perform this adaptation and customisation manually is an expensive and time-consuming task. Additionally, as devices become increasingly interconnected, and as users are able to switch from one to another at will, it is essential to avoid confusing and disorienting the user (Chincholle, 2000). To this end, it is desirable to keep the content and structure of a document constant while optimising the presentation for each device to ensure usability. This is especially true for small or mobile devices. Figure 1 illustrates this issue. The screen on the left shows a report that might have been dynamically generated for a PC based web browser in response to a query about “financial threats to the Russian space program”. This report integrates information retrieved from multiple sources, and the information is structured and organised in order to be easily understandable. In particular, each piece of information is labelled with its role in the report. The most important part of the report is shown in bold at the very top. Then there is elaboration information, which gives more information about this first sentence. This is followed with information indicating the likely impact (consequence). The right hand side of Figure 1 shows the presentation of the same information, now delivered on a mobile phone. Given the different affordability of the delivery channel, the presentation and, in particular, the navigation is quite different, while the content and structure of the information being presented is constant. Our approach to ensuring coherence and consistency of information across devices is to use Myriad (Paris et al., 2004), a platform for contextualised information retrieval and delivery based on theories and techniques from natural language generation. Our system produces virtual documents …

Content truncated to 500 words