INFORMATION-RETRIEVAL, IMAGING AND PROBABILISTIC LOGIC

Imaging is a class of non-Bayesian methods for the revision of probabi lity density functions originally proposed as a semantics for conditio nal logic. Two of these revision functions, standard imaging and gener al imaging, have successfully been applied to modelling information re trieval by Crestani and van Rijsbergen. Due to the problematic nature of a ''direct'' implementation of imaging-revision functions, in this paper we propose their alternative implementation by representing the semantic structure that underlies imaging-based conditional logics in the language of a probabilistic (Bayesian) logic. Besides showing the potential of this ''Bayesian'' tool for the representation of non-Baye sian revision functions, recasting these models of information retriev al in such a general purpose knowledge representation and reasoning to ol paves the way to a possible integration of these models with other more I(R-oriented models of IR, and to the exploitation of general-pur pose domain-knowledge.