タイトル無し

概要

論文の詳細を見る
Effective search strategies for Randomized Controlled Trials (RCT) in a Japanese medical database are investigated by the application of a relevance-feedback information retrieval model. A training set and four test sets were prepared from the bibliographic references in JMEDICINE. The references in two of the test sets were taken from the same journal the training set was taken from and the other two from different journals. Precision values assigned to the terms in title, abstract and subject index fields in the training set were used to construct probabilistic and Boolean search strategies. For the same-journal test sets, both strategies were well performed with the title and subject index fields but neither search of the different-journal test sets worked well. The reasons are discussed, and the measures to make RCT searches more effective are proposed in terms of searching, indexing, and making abstracts.
日本薬学図書館協議会の論文