Research
My research interests lie within the use of Knowledge by distributed autonomous systems, or Agents. These agents may provide specific support for an individual (for example, Agentic AI), or within an autonomous community of agents (Multi-Agent Systems). Much of my focus has been on how formal, symbolic knowledge is defined and shared between agents, thus allowing collaboration and co-operation, using a variety of AI based techniques, through heuristic search, dialogue and negotiation, machine learning (neurosymbolic) and more recently, through the exploitation of GAI (generative AI) techniques.
Service Discovery for Multi-Agent Systems
For the last 25 years, much of my research has focused on the use of ontological knowledge by agents; specifically, to support service discovery, provision and use. Much of my earlier work in this area (through DARPA's CoABS project) included the development of the OWL-S ontologies with the other members of the OWL-S coalition, as well as the use of the OWL-S profile for service discovery. However, my work has also explored the use of service provision and agent self-organization. I’ve also explored the use of service discovery and data harmonisation for Semantic Web Services within Grid environments. My other research interests include the use of agents in pervasive systems, from discovery of devices, through to the use of awareness for service provision (as illustrated in the BluScreen system for public signage).
Knowledge-Based Laboratory Environments
The use of services and workflows have been fundamental to describing processes within business services and laboratory environments. My involvement was grounded in this area from the days of Kesselman & Foster's Computational Grid, used to support larger scale, decentralised scientific experiments across different institutions. This included my work on CombiChem & MyGrid projects, looking at the provision of services on physical devices and bio-informatics services, and more recently examining the support of autonomous services within modern labs through LIMS environments and schema representations.
Much of my recent work with colleagues here at Liverpool has explored the exploitation of GAI for ontology engineering, specifically within the realm of scientific workflows; exploring for example schematic representations such as AnIML and how they can be modelled semantically (e.g. using OWL). As part of the Lab headed by Dr Tamma, we explored not only how GAI can be used to map knowledge-poor schemata into semantically rich representations, but also in decomposing and retrofitting Competency Questions that can be used as the basis for engineering sharable and evolvable ontologies.
AI in Education
A more recent direction has involved the use of formal knowledge models (ontologies) and Knowledge Graphs that describe domains for use in the modelling of pedagogical material. Through work with Dr AlKhuzaey (and colleagues here in Liverpool), we have explored the notion of item difficulty - i.e. the pedagogical or cognitive challenge that questions pose to students, and proposed mechanisms by which questions that assess a greater understanding of pedagogical material can be generated, through the use of formal ontological and knowledge modelling approaches.
Research collaborations
Valentina Tamma
Exploring the use of decentralised negotiation techniques (including Argumentation) to facilitate agreement over ontological concepts to support communication
Samah AlKhuzaey
Umm Al-Qura University
Investigation of Item Difficulty and Question Generation using formal ontological Models.