The development of 'for experts systems' as heuristic reasoning platforms in risk decision support: a consideration of tool design, technology transfer and compatability with Bayesian decision analysis
Arthur, J. G. (2007) The development of 'for experts systems' as heuristic reasoning platforms in risk decision support: a consideration of tool design, technology transfer and compatability with Bayesian decision analysis. PhD thesis, University of Warwick.
WRAP_THESIS_Arthur_2007.pdf - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Official URL: http://webcat.warwick.ac.uk/record=b2241736~S15
This work considers the creation of two risk and decision support systems, one for the National Air Traffic Services of the UK and one for Unilever, a multi-national. Their development contributes to risk decision science in the area of decision support in particular. This contribution is based on the development real-life systems, it has three key elements. One, it addresses the fact that, for practical environments like these, the science of risk and decisions is insufficiently resolved to be accepted and easily used. Two, the systems share an arena with subjective Bayesian decision analysis. The benefits of a hybrid form of the two approaches to generate higher levels of user acceptance and organisational transfer is discussed. Three, they take the unique approach of being 'for experts' systems rather than 'expert systems'. This approach offers a number of benefits to applied user communities. These include: a decision support system which remains grounded within the reasoning world view of the decision makers; an expansion and refinement of the existing 'natural heuristics' that decision makers use currently; a scoring and visualisation environment which is both fast and flexible but allows for, previously unavailable, levels of reasoning transparency and comparison. Taken in total the combination of the tool design, the heuristic artefacts within them and their influence on the hosts organisations, the two systems have proven they can provide an effective and valued 'heuristic reasoning platform' for risks and issues. A future research direction is to explore ways in which the highly transferable heuristic artefacts in these systems, particularly measurement and data manipulation, might be strengthened via hybridisation with more powerful, but less transferred, formal systems like Bayes decision analysis.
|Item Type:||Thesis or Dissertation (PhD)|
|Subjects:||Q Science > QA Mathematics|
|Library of Congress Subject Headings (LCSH):||Bayesian statistical decision theory, Decision making -- Mathematical models, Risk assessment -- Mathematical models, Heuristic programming, Programming (Mathematics)|
|Official Date:||December 2007|
|Institution:||University of Warwick|
|Theses Department:||Department of Statistics|
|Supervisor(s)/Advisor:||Smith, J. Q., 1953-|
|Format of File:|
|Extent:||326 leaves : charts|
Actions (login required)