Probabilistic Retrieval Revisited

Open Access
Publisher Website
Google Scholar
Abstract
The well-known probabilistic model for information retrieval based on Bayesian conditioning of probability functions is examined. It is extended to allow conditioning based on evidence derived from the ‘passage of experience’ which may be non-propositional in nature. This latter form of conditioning is derived from Jeffrey's work on probability kinematics and it is compared with the Dempster–Shafer approach to revising belief functions whilst motivating its appropriateness for adaptive information retrieval. This new form of conditioning is combined with a non-classical logic to define a new probabilistic model for information retrieval.

This publication has 0 references indexed in Scilit: