作者: James P Callan , W.Bruce Croft , John Broglio
DOI: 10.1016/0306-4573(94)00050-D
关键词: Systems design 、 Information needs 、 Bayesian inference 、 Statistical model 、 Information retrieval 、 Network model 、 Document retrieval 、 Relevance feedback 、 Query expansion 、 Computer science 、 Artificial intelligence
摘要: Abstract INQUERY is a probabilistic information retrieval system based upon Bayesian inference network model. This paper describes recent improvements to the as result of participation in TIPSTER project and TREC-2 conference. Improvements include transforming forms-based specifications needs into complex structured queries, automatic query expansion, recognition features documents, relevance feedback, simulated document routing. Experiments with one- two-gigabyte collections are also described.