北京大学本科生业论文-自动标注汉语多音词.doc

本科生毕业论文 题目： 自动标注汉语多音词  Automatic Tagging of Chinese Polyphony Word and Related Research 姓 名： 何辉辉 学 号： 院 系： 信息科学技术学院 专 业： 计算机科学与技术 指导教师： 段慧明 二〇一一年六月一日 摘要 汉语中存在着一字多音现象，很多高频字都具有两种或更多的读音，而这些多音字的读音并没有统一的规则可循。汉语多音词的自动标注，即将汉字序列转换成相应的拼音序列，是语音合成(TTS)系统的关键步骤。由于汉语文化的多义性和上下文相关性等原因，目前的汉语注音系统很难达到100%的正确率。 本文研究了多音词的读音规律，总结了中文分词和词性标记的相关技术；设计了汉语多音词的注音，研究词性标记串的注音问题 关键词：汉语多音词；词性标记；自动注音；统计与规则。 Automatic Tagging of Chinese Polyphony Word and Related Research Abstract In Chinese, there are many frequently used characters, which have more than one pronunciation, and how to automatically tag those characters correctly is a key step in TTS system. However, because the pronunciation of a certain character is related to its context and may have ambiguity, its very hard to find a universal rule. As a result, current systems can hardly achieve a 100% precision. This thesis researched the law on Chinese polyphone words’ pronunciation, and summarized the related techniques on Chinese segmentation and POS tagging. In our work, we designed a pinyin tagging system, which was designed by Java and MySQL database, to focus on the problem after POS tagging. We proposed an algorithm that combined the traditional statistical method and heuristic rules to achieve a better result. Moreover, we also conducted extensive experiments, and the result showed that it can achieve accuracy of 95.2%; with the improvement of word database, its accuracy will continually increase. Key words： Automatic Tagging; Statistics and Rules. 目录 摘要 2 Abstract 3 第一章 绪论 5 1.1 问题背景和研究意义 5 1.2 发展现状 6 1.3 本文组织结构 6 第二章 读音 7 2.1 多音字与多音词 7 2.2 读音分类 7 2.3 读音规律 8 第三章 分词与词性标注 10 3.1 分词定义 10 3.2 现有分词技术 10 3.3 N元模型 11 3.4 词性标注定义 12 3.5 词性标记技术 12 3.6 隐马尔可夫模型 13 第四章 拼音标注 15 4.1 标音定义 15 4.2 现有拼音标注技术 15 4.3 规则与统计相结合的注音方法 16 第五章 多音词注音系统的实现 17 5.1 注音系统简介 17 5.2 开发环境 17 5.3 数据库表结构 17 5.4 程序模块结构 18 5.5 注音流程 20 5.6 功能测试