Invited Speakers

Carole Goble, School of Computer Science, University of Manchester
Jon Oberlander, School of Informatics, University of Edinburgh
Judy Kay, School of Information Technology, University of Sydney

Abstracts

Web + Semantic Web enables Adaptive Hypermedia? - Carole Goble

The Web is the most successful hypertext ever. It is only natural and appropriate that hypertext research, and its researchers, would thus adapt to the Web and its ways, despite the fact that the underlying simple hypertext infrastructure’s theoretical foundation is simple (good) and limited (bad). The model is based almost entirely around nodes with links playing second fiddle - embedded and difficult to author, maintain, share and adapt.

The Semantic Web is based on the notion of exposing metadata about resources in an explicit, machine-processable way. By doing so, we open up the possibility of using machine processing in order to help us search, organize and understand our data. So far this has largely been used to help in provide more effective search, describing web services and driving applications like enterprise integration.

Potentially, Semantic Web technologies could be used to adapt and evolve the embedded link structure of the Web, effectively building a semantically-driven dynamic open hypermedia system. Additional metadata and reasoning components add links dynamically between resources, potentially improving the linking and navigation of Web pages intended for end-users. Multiple knowledge resources can be used to personalise links, providing better relations, presentations and links that are more relevant to users. Systems such as COHSE and Magpie have pioneered these ideas, and can be presented as allies to the notions of semantic wikis and social tagging. Thus they bridge between the high-brow world of Semantic Web as perceived by the Artificial Intelligence community, the low-brow world of “collective intelligence” as perceived by the Web 2.0 community, the Web designers’ world of content and link creation and the users’ experience of hypermedia navigation.

The talk highlights relevant aspects of Semantic Web technologies and discusses how these can contribute to the construction of links, using examples from the COHSE system. We address whether, with the help of the Semantic Web, the Web can learn to adapt to hypermedia design and hypermedia users.

Adapting NLP to Adaptive Hypermedia - Jon Oberlander

Natural Language Processing (NLP) techniques ought to be really useful for people building adaptive hypermedia (AH) systems. This talk explores the gap between theory and practice, illustrating it with examples of things that do (and don't work), and it suggests a way of closing the gap. In theory, NLP sub-systems should help find, filter and format information for re-presentation in AH systems. So there ought to be lots of cross-fertilisation between NLP and AH. It is true that some projects have effectively brought them together; particularly on the formatting---or information presentation---side, natural language generation systems have allowed quite fine-grained personalisation of information to the language, interests and history of individual users. But in practice, NLP has been less useful to AH than one might have expected. Now, one reason for this is that the information to be presented has to come from somewhere, and NLP support for AH authors is not as good as it should be. Arguably, where NLP could really make a difference is on the finding and filtering side. State-of-the-art information extraction tools can increase author productivity, and help make fine-grained personalisation more practical.

Scrutable adaptation: because we can and must - Judy Kay

Beginning with the motivations for scrutability, the talk introduces PLUS, a vision of Pervasive Lifelong User-models that are Scrutable. The foundation for PLUS is the Accretion/Resolution representation which is at the base of an architecture for active user models that can drive adaptive hypermedia, with support for scrutability. Essentially, it aims to be the simplest representation that enables user control over use and interpretation of their user model, while also supporting reasoning under the uncertainty that goes with user modelling.

The core of the talk illustrates PLUS in terms of its existing, implemented elements of this vision. These are mainly derived from my work: the Personis-lite version of Accretion/Resolution, and the Personis-plus active user models; tools like VCM to elicit conceptual understanding and knowledge; SIV as an example of an interface that will be essential to support scrutability of large user models; a framework for defining user model ontologies and distributing them, managing movement of partial user models in ubiquitous computing; scrutability-preserving service discovery for pervasive computing; and taking scrutability beyond the user model in scrutably adaptive hypertext.

To illustrate how this notion of scrutability impacts an adaptive hypertext application, I will present some examples: the Just-in-time Training system (JITT), the Unix tutor with scrutably adaptive hypermedia and MyPlace, a pervasive personalised service that informs the user about relevant people, things and services.

The concluding section is a research agenda for some of the essential elements of this PLUS vision, with the role, challenges and trade-offs in scrutability for adaptive hypermedia.