Robert Stevens Research

Bio-Ontologies

What I consider to be the background for ontology can be summarised as an attempt to formalise the knowledge component of biology and make it computationally amenable. The aim of computational use of knowledge has led me to use description logics, particularly the Web Ontology Language (OWL), as a means of representing these ontologies. In this field, my research interests include, but are not limited to:

• Building ontologies that can be used in applications that allow biological information to be queried and analysed.
• The process of developing ontologies; the methodologies used and, in particular, the distinctions that it is useful to make.
• the tools used, especially through and involvement with the development of the Protege Development Environment in its OWL incarnation. >
• The broader social aspect of ontology building, including the issues involved in collaborative ontology building and the role of ontology comprehension in that endeavour.
• The use of ontology design patterns that make modelling easier, more consistent and generative.
• the use of text to help support the rapid start of ontology development.
• All these come together to help make rich ontologies that make descriptions of the knowledge as useful as possible, rather than simply a collection of labels for concepts.

Making the descriptive aspect of biological data more computationally amenable goes hand in hand with my research interests in e-Science. These primarily focus on the use of workflows to capture bioinformatics protocols through the Taverna Workflow Workbench. Workflows capture scientific protocols at a a high level and make them explicit. Workflow also enable the provenance of an an arbitrary experiment to be captured and exploited in a systematic, consitent manner. Here, the semantic, knowledge components of bioinformatics come into play again. Describing workflows, the services they use and the data they generate all require semantic descriptions that are enabled by ontologies.

Accessibility

Computers have been a tremendous boon to visually disabled people. The proliferation of electronic information and the means to process that information has opened up a broad range of opportunities to visually disabled people that were hereto closed. Screenreaders, either through braille or synthetic speech, make this information available to visually disabled people.

Unfortunately, simply sticking a speech synthesiser onto a computer does not make the applications or information they provide accessible or useful. Web pages, for example, can be highly complex information artefacts. Mathematics, diagrams, music and science notations all have complexities over and above that seen in plain text. It is providing useful access to these kinds of complex information that is the overall aim of my research in this area.

Underpinning my approach to this topic is an attempt to characterise the problems that non-visual interaction with icomplex information presents. By understanding why a visual presentation works for sighted people we can often make inferences about what provision we need to give in non-visual interaction. So, we often ask questions such as `what is it about node and arc diagrams that make them work?"; `what is a glance, what information does it supply?". If such questions can be answered, then the possibility of designing a good acces solution is far higher.