FULL-TEXT RETRIEVAL OF MEDICAL JOURNAL LITERATURE

我们目前的研究表明，搜索 医学期刊一贯产生的百分比要高得多 相关文件，而不是搜索他们的书目代理。 它还表明，全文搜索会产生显著的 更多不相关的项目。稳步增长的数字 全文生物医学数据库，包括在线和光盘 格式，对更好的搜索技术进行了研究 这些数据库是医学研究的重要领域 信息科学。我们建议进行这样一项 调查。通过创建和分析文件的详细记录 在两个支持完整的大型数据库的系统上的操作 文本生物医学期刊，我们将开发搜索技术 在不显著丢失文档的情况下删除不相关的文档 这些都是相关的。我们将特别注意 同义词聚合的过程，对检索非常重要 全文档中的相关文件。通过分析技术 在一个系统上成功，但在另一个系统上不成功 软件和文件结构，我们将描述这些设计 最有助于在全文中成功搜索的特性 数据库。

Our current research demonstrates that searching the full-text of medical journals consistently yields a much higher percentage of relevant documents than searching their bibliographic surrogates. It also shows that full-text searches result in a significantly greater number of irrelevant items. A steadily increasing number of full-text biomedical databases, in both online and compact disk format, makes an investigation of better searching techniques for these databases an important area of research in medical information science. We propose to undertake such an investigation. By creating and analyzing careful records of file manipulations on two systems which support large databases of full- text biomedical journals, we shall develop searching techniques for eliminating irrelevant documents without significant loss of those which are relevant. We shall pay particular attention to the process of synonym aggregation, so important for retrieval of relevant documents in full-text files. By analyzing techniques successful on one system but not on the other in terms of system software and file structure, we shall delineate those design features most conducive to successful searching in the full-text databases.