摘要：First it was the mini-car, then came the miniskirt, and even the mini-puter; now we have the mini-abstract. Such, at least, is the term invented by Mrs. Lunin in her paper, "The development of a machine-searchable index-abstract and its application to biomedical literature." These abstracts consist of keywords arranged in a syntactic structure, any number of which can be used for an article. The particular project which Mrs. Lunin was working on had to do with the relationship between trace metals in the body and neoplasms; the field, therefore, is a concise one, and the material homogeneous. There appear to be two basic syntactic structures possible; the first deals with the amount of any element in any part of the body under varying circumstances, and the second with the method of determining the presence of that element. Because of this reduction to a small number of basic types, vocabulary control seems to be quite simple. Mrs. Lunin suggests that it would be possible to extend this method of abstracts to other areas of the sciences, and indeed this may well be so in other equally welldefined areas, but I would be particularly interested to find out if its application to a general medical library would be defeated by the proliferation of syntactic structures. Another disadvantage of this system is the number of abstracts which have to be written for each article. For example, the presence of several trace elements in several organs of the body would require that many abstracts, the appropriate terms differing but using the same basic syntax. Presumably in an article relating to trace elements and other things, this system would neglect everything not about trace elements; in the general library, this would not be possible, and the number of abstracts written would increase correspondingly. However, for the project inhand, the system seems to work extremely well; the information produced is easily accessed because of the extreme formalism, an advantage which probably makes up for the loss of information to be found in the traditional abstract, which has the disadvantage of having to be humanly read. I am not sure, however, why some of the keywords have to be entered as abbreviations; the system has gone far in the direction of plain language that it seems a pity to go back to indexers' shorthand. This would seem to be a stumbling-block against the day when the researcher himself might try to consult the system.
