Library and Information Science

Generally speaking, the document retrieval operation is based on probabilistic conversion and matching of both indexing and query languages. Retrieval efficiency can be improved by the method of trial and error at the man/machine interface on on-line retrieval systems. The advantages are: 1) remote access by the requester in person, 2) relatively free conversation with the file, 3) shorter wait times, 4) continuous updating of the file, and 5) updating from remote stations, as the case may be. Examples of existing systems such as NASA/DIALOG, Japan Information Processing Development Center's JOLDOR, MIT/TIP, and MIT/Intrex are described, with particular emphasis on file structure and search procedure of DIALOG. Shortcomings of these systems at the moment are longer search times, which will be much improved by time sharing and vocabulary control.