- A Study of Natural Language Processing for Indexing -- A Statistical Survey of Medical Terminology--
- No.5, p.51-72
In relation with the problem of language processing for indexing and searching purposes, the writer carried out an experimental research of the processing of natural language in documents. The field of medicine was selected as the research subject, and the samples were taken from words appeared in titles of articles in the area of pathology.
In extracting terms from natural language, it is required to analyze the agglutinativeness of Chinese characters used in Japanese language. A Chinese character, as is well known, has its own meaning just like a word has. Consequently, the Chinese characters can be said to have word-construction function through which their different associations become to indicate different concepts. This causes a difficulty in identifying word division or unit word which is basic to the study of terms. Further, this is resulted in the ambiguity of structure expressed by the association forms of unit words.
To make clear the characteristics mentioned above, quantitative measurement by the statistical method may be of help. In this article, the writer tried to clarify, by this method, the nature of medical terminology and to evaluate its value as indexing terminology.
(Graduate Student, Japan Library School)
- 本文PDF (2,631K)