A combination of expert opinion approach to probabilistic information retrieval, part 1: The conceptual model

摘要

This paper presents a new version of Robertson, Maron, and Cooper's probabilistic information retrieval model. Model 3. The new version of Model 3, based on the statistical technique of the combination of expert opinion, attempts to overcome difficulties with the previous version of Model 3: Those faced by most probabilistic retrieval models including various independence and small sample problems and those specific to Model 3. Model 3 seeks to combine two earlier models of retrieval, but how to do so is unresolved. The present model shows how that combination can occur. Furthermore, by advocating the use of indexer and user subjective probabilities and determination of dependent terms, it shows a way of dealing with the problems of small samples and independence/dependence of index terms.