My principal research interest is in method engineering and method automation, focusing on the elicitation, modeling and analysis of knowledge that human experts apply in problem solving and decision making, the engineering of ontologies and processes capturing that knowledge, and the automation of the said processes.
This interest falls within the various research fields concerned with the transfer, preservation and automation of knowledge.
Study of problem solving and decision making knowledge of human experts underlines numerous research problems and high-impact applications. For example:
- How do we elicit, represent and automate methods of human experts and make them available to millions in interactive format (which is not the case for web pages, books, scientific publications, and more generally, text, images and video)?
- How do we transform an expert’s method into algorithms which can automatically deliver recommendations/advice at a scale that is elusive to the expert?
- How do we evaluate recommendation algorithms against methods of human experts(e.g., trading algorithms, fraud detection algorithms)?
- How do we evaluate the quality of an expert’s problem solving and decision making independently from their reputation?
- How do we evaluate the quality of existing systems engineering methods, and how do we create new systematic methods based on expert engineer’s knowledge?
- How do we create methods for incorporating into Business Intelligence systems the knowledge of expert decision makers, so that the systems can deliver advice?
This general research interest translates into the following research goals:
- To construct, refine and empirically test novel conceptualizations, mathematical models and methods for the elicitation, modeling and analysis of knowledge that underlies problem solving and decision making methods, specifically methods applicable to settings characterized by unavailable/incomplete quantitative estimates of probability and/or utility and the availability of variously imprecise, vague, incomplete, conflicting, and unstable/changing qualitative decision information and advice from many stakeholders. Examples include methods applied to deliver complex services, such as an engineer applies when designing an information system, a sports coach when advising an athlete on how to train, an architect when designing an airport, and so on.
- Apply conceptualizations, models and methods to method engineering in specific domains and empirically evaluate the relevance of the resulting methods through industry transfer, as I believe that the empirical evaluation of method engineering and method automation needs to be performed under industry-specific resource constraints.
- Form an interdisciplinary research group focused on the elicitation, modeling and analysis of knowledge that underlies problem solving and decision making methods.
- Create a graduate study program specializing computer science students for the elicitation, modeling and analysis of problem solving and decision making knowledge, preparing them thereby to lead the engineering of future decision-support and recommendation systems.
I believe that the success of computer systems which aim to automate problem solving and decision making methods of human experts depends to an important extent on how well the expert’s knowledge is elicited, understood and modeled. Based on my research experience, I also believe that it is not relevant to search for universal conceptualizations, models and methods for knowledge elicitation, modeling and automation, but instead work on domain-specific conceptualizations, models and methods, and when possible, abstract from these the method engineering guidelines which systematically reappear as relevant across domains.