##
**Subjective probability versus belief functions in artificial intelligence.**
*(English)*
Zbl 0795.60002

Summary: We maintain that probability should be preferred with respect to other existing theories as an inferential tool to cope with uncertain knowledge (and not only in the field of Artificial Intelligence). In particular, we discuss probability versus belief functions, by referring to an “abstract” formulation of a classical example. An important aspect is that there is no need to distinguish available evidence from any other possible assumption, which can be looked upon as a potential evidence: if the conditional probability (or Bayesian) approach is set up taking into account all the relevant aspects, i.e. if there is no beforehand given structure on the set of events (which in turns are singled-out by propositions) and if probability is interpreted as degree of belief, then the Bayesian approach leads to very general results that fit (in particular) with those obtained by the belief function model.

### MSC:

60A99 | Foundations of probability theory |

### Keywords:

subjective probability; potential evidence; conditional probability; degree of belief; Bayesian approach; belief function model
PDF
BibTeX
XML
Cite

\textit{R. Scozzafava}, Int. J. Gen. Syst. 22, No. 2, 197--206 (1994; Zbl 0795.60002)

Full Text:
DOI

### References:

[1] | Coletti G., Lecture Notes in Computer Science 521 pp 106– (1991) |

[2] | Gilio A., Atti dette giornate A1RO su Ricerca Operativa e Intelligenza Artificiale pp 317– (1988) |

[3] | Gilio A., Bollettino Unione Matematica Italiana pp 645– (1990) |

[4] | Scozzafava R., International School of Mathematics pp 244– (1991) · Zbl 0811.60004 |

[5] | Scozzafava R., Proc. IPMU 1992, (Palma de Mallorca), Uncertainty in Intelligent Systems (1992) |

[6] | Smets P., Fuzzy computing pp 17– (1988) |

This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.