Abstract
The possibility of the text categorization using radicals of Kanji (bushu) has been examined in Japanese texts. Kanjis in the texts were converted into the corresponding radicals. The frequencies of the radicals, as well as frequency orders are acquired with different subject categories of texts. The same tendencies were observed for the titles and for the full text of the articles. The "Top 7 Radicals" frequented in all the categories, their order however differed by categories. The typically frequent radicals related to the subject fields were observed in subcategories of industries. The frequency orders of radicals were stable within a journal, when analyzed by whole issues of scientific journals, whereas it varied when analyzed by individual articles. The radicals seem to be applicable to the bulky text categorization.
