In conjunction with Weiru Liu
(Queen's University of Belfast)
we have extend the fusion rules approach to modelling and merging uncertain
information that is defined either on textentries, or at different levels
of granularities of textentries, as well as to modelling and reasoning
with structured reports that
contain semantic heterogeneous
uncertain information on more complex elements in structured reports (such as uncertainty distributions on subtrees).
Representing and reasoning with uncertainty in information is potentially very useful for key applications of the fusion rules approach. It allows handling of both object-level and meta-level uncertainty. We have harnessed a range of theoretical approaches to uncertainty including probabilisitc, possibilistic, and belief function approaches. This has included techniques for translating between different forms of uncertainty, and for dealing with different degrees of granularity. We have also harnessed and extended techniques for merging uncertain information including interval-based techniques and Dempster's rule of combination. In addition, we have extended techniques for evaluating the degree of conflict between sources of uncertain information. So far our work is covered in the following papers.