Abstract
This paper describes a general assignment method for combining evidences in the form of mass assignments over the power set, P (X), of a set of labels. It also describes an iterative assignment method for updating an apriori assignment over P (X) with evidences E1, ..., Er all expressed as assignments over P (X). This can be thought of as an extension of Bayesian updating with uncertain information. A minimum relative information optimisation of the updated assignment relative to the apriori is used. The methods have application to knowledge engineering for processing rules with uncertainties, causal nets and other representations for inference purposes. Solutions to the non-monotonic logic and abduction problems are special cases of this inference process. The methods are not the same as the Dempster Shafer theory of evidential reasoning, [Shafer 1976] but does use Shafer's belief function form of representing uncertainty.
Preview
Unable to display preview. Download preview PDF.
8. References
BALDWIN J. F. (To appear 1990) "Combining Evidences for evidential reasoning", Int. J of Intelligent Systems, pp 1–40
BALDWIN J. F. (1990), I.T.R.C Report, University of Bristol, “Assignment Methods for Evidential Reasoning for Knowledge Engineering”, pp 1–45.
BALDWIN J. F., PILSWORTH, B., MARTIN, T., (1987), “FRIL Manual”, Fril Systems Ltd, St Anne's House, St Anne's Rd, Bristol BS4 4A, UK
SHAFER G., (1976), “A mathematical theory of evidence”, Princeton Univ. Press
ZADEH L., (1965), “Fuzzy sets”, Information and Control, 8, pp 338–353.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1991 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Baldwin, J.F. (1991). Towards a general theory of evidential reasoning. In: Bouchon-Meunier, B., Yager, R.R., Zadeh, L.A. (eds) Uncertainty in Knowledge Bases. IPMU 1990. Lecture Notes in Computer Science, vol 521. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0028122
Download citation
DOI: https://doi.org/10.1007/BFb0028122
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-54346-6
Online ISBN: 978-3-540-47580-4
eBook Packages: Springer Book Archive