一种语句级汉字输入技术的研究

A Sentence-Level Chinese Character Input Method

提出了一种语句级汉字输入技术,把Ngram统计语言模型与语言规则结合起来共同指导拼音流到文字流的转换过程。描述了拼音流自动切分、词网格生成、最优语句候选搜索和系统的自适应学习机制等技术。该语句级汉字输入技术的音字转换正确率达9028%,目前它已用于Microsoft公司的微软拼音输入法中。

A sentence level Chinese character input method is proposed, which integrates N gram statistical language model and linguistic rules to guide Chinese syllable text conversion Chinese syllable stream segmentation, word lattice formation, the optimal sentence hypothesis search and self adaptive learning mechanism are described The syllable text conversion accuracy of 90.28% is achieved The technique has been applied in Microsoft Cooperation’s Chinese Pinyin input method