股票领域中的一种中文问句分类方法

A Classification of Chinese Questions in the Domain of Stocks

摘要: 问句分类是问答系统的重要组成部分，分类的准确性在一定层面上决定了答案抽取的策略和准确性。本文所介绍的分类方法是以网络论坛中的用户提问作为样本集，并分析每种问题所含的比例，有侧重地提出一套适合于股票领域中问句分类的规则。系统将该领域问题分为概念、投资家、个股三大类。其中，概念类中又包括定义类1、定义类2、比较类、描述类、原因类、列举类、属性类等七类。初步测试表明这种分类方法有效地提高了股票领域问句分类的准确性。

Abstract: Question classification is a necessary part of Q & A system, and it determines the accuracy of ans- wer extracting in some degree. This paper introduces a classification that depends on the sample set including questions from stock forum. We divide questions into three main parts: conception, investors and stocks. And we also divide conception class into seven parts: definition I, definition II, comparison, description, reason, enumeration, and attribute. The test results show that this approach did good performance in stock domain.