智能型汉语语句生成和诊断系统及其机器学习问题

智能型汉语语句生成和诊断系统必须建立在完善的知识库（包括词库、规则库和学生错误模型，特别是规则库）的基础之上，但是完善的知识库又不可能－蹴而就，因此重要的手段之一就是是靠系统本射的机制来逐渐学习完善，即系统应具有可扩充性。本文讨论了智能型汉语语名生成和诊断系统中可扩充性的实现方法，即机器学习问题，主要介绍了系统中自学机制。由于系统正处于开发研制阶段，因此各个方面都有特进一步的完善。

An Intelligent System for generation and Diagnosing of Chinese sentences and its Machine Learning Issue

The construction of Intellectual System of Sentence Producting and Diagnosing should be based on a complete knowledge-base including word base, rule base and student mistake model. Since it is impossible for a complete knowledge-base to be established in a short term, it should be formed step by step. One of the important way to do this is to get the system improved by itself, that is to say, the system should be expansible.It is discussed in this paper the way to realize this expansibility, or machine learning.The self-learning mechanism of this system is introduced here. Since it is at the beginning stage of the study, the system presented here is to be improved later on.