Computer Science

Update - 29/01/06
I realise this page has been updated infrequently for which I apologise. I will now attempt to keep it up to date as much as is possible.

Accessibility

I am a "Participant in Good Standing" (or PiGs as Joe Clark likes to call us) of the Web Content Accessibility Guidelines (WCAG) working group. WCAG WG is part of the World Wide Web Consortium's (W3C) Web Accessibility Initiative (WAI). WCAG 2.0 is moving along, we made some really good progress at the recent face to face meeting in Dublin. One of the key points to come out of the meeting was the strong possibility of using an assumed baseline of UAAG 1.0. While we accept that right now no user agent meets this guideline we also felt it was not only unfair to place sole onus on web authors, but also ineffective at combating accessibility issues.

Pending Publication: WCAG 2.0 General Techniques

Semantic Web

My current research in Semantic Web (SW) technology tends to be around three axes. These are

On the SW at present there are large amount of objects which encapsulate complex data. The SW provides both benefits and hindrances to the identification of these objects. While the structure of the SW provides large amounts of extensible properties about objects, adding a real world richness to them, it is also currently very binary. More traditional fields which deal with data have already found ways to address this, and I wish to bring some of these solutions to the SW to remove some of the very arbitrary nature of matching. I feel this will be increasingly important as the SW grows. With the development of more sophisticated inferencing it will be critical to ensure that the inevitable degradation of data quality inherent with the uptake of a technology does not detract from the strengths which we value the SW for.

Trust is also a key issue on the SW. It is important that we recognise that most SW applications are currently running in clean room environments. They receive very specific data, not in terms of syntactic correctness, but in terms of information correctness. Thus many of the application have a certain amount of naivety regarding data they receive, or the source which they receive it. Ensuring that applications can identify, or at the recognise and sandbox unidentified, data is an important step towards making the SW robust enough to cope with the real world.

Inferencing is to a large extent an AI topic it's clear roots in such things as expert systems. However with the large domain of information available on the SW it is becoming an increasingly attractive prospect to tap these resources. The flexibility of SW technology also allows us to model data of a system to a granularity which is significantly more difficult in more traditional mediums. This specificity of systems modeling can allow very simple user input to create a number of indirect implications. I am keen to explore the use of simple user metrics to act as a catalyst for inferencing. Finding critical areas of assessment can allow minimal user input to combine with numerous inferences made on the part of the system to provide new insight. The beginning of such a system can be seen in the paper I presented at the first FOAF conference in Galway, Ireland. Although this paper outlined such as system, it did so on a very specific topic, which was deliberately constrained.

Publications: A model of trust and anonymity in a content rating system for e-learning systems, (2004), Proceedings of the 1st Workshop on Friend of a Friend, Social Networking and the Semantic Web.

Unpublished: Situation and Identity: A generalisation of Inverse Functional Properties, with Joe Geldart.

Human Computer Interaction

As part of my internship at the University of Sunderland I worked in the Human Computer Systems Group (HCSG). HCSG are one of the leading HCI labs in the country. They were presented with the award for best paper at HCI 2003 for the introduction of their DARE model. DARE is a system for critical evaluating usability heuristic methodologies. During my time at HCSG, which to some extent ongoing while I remain at the University of Sunderland, I was working on a paper looking at errors on the web. As the research stands it presents the most complete (to my knowledge) model of errors and their occurrence on the web. I am currently working on a user centred model using generative transition networks and a set of experiments to explore some of the postulations formed from the research.