中文文献自动分类的理论与实践

AUTOMATIC CLASSIFICATION OF DOCUMENTS IN CHINESE— THEORY AND PRACTICE

本文介绍的中文文献自动分类实践，是基于文献主题属性的分析，运用概率标引技术和Bayes分类准则等理论依据，防人工实际工作过程而实现的一种仿人算法。系统使用了以加权的题中关键词为基础的切实可行的方法，借助中文文献自动抽词系统的技术成果，以期达到科学性、继承性、实用性兼顿的目的。算法在微型机IBM-5550上实现。

The practices in automatic classification in Chinese introduced in this paper are based on an analysis of the theoretical criteria for the attributes of document titles employing the probability indexing technique and Bayest classification norms forming a kind of man-micmicking algorithm realised in imitating the actual process of manual work. The system makes use of the really feasible method taking the keywords in the weighted topics as the basis aided by the technological results of the automatic term-extraction system in Chinese with the expectation of achieving the purpose of being seientitic, inheritable and applicable. The algorithm is realised on the microcomputer IBM-5550.