A combination of expert opinion approach to probabilistic information retrieval, part 2: Mathematical treatment of CEO model 3

摘要

This paper fills in mathematical details of a combination of expert opinion approach to probabilistic information retrieval, described conceptually in part 1. Included are: how beta distributions used for indexer and system individual term distributions are updated and transformed; how in Model 1 indexers' individual term distributions are calibrated based on past performance; how individual term distributions are combined to give Models 1 and 2 distributions; how Models 1 and 2 distributions are combined to yield the Model 3 distribution; how the odds formula version of Robertson, Maron, and Cooper's Model 3 can be interpreted as a special case of the combination of expert opinion version of Model 3.