摘要
音节切分是整句拼音转换的基础,由于拼音的特殊性,存在歧义切分的可能。如果采用最少分词算法只能得到一种切分结果,不能保证整句拼音转换的正确性。提出一种音节切分算法,通过插入音素节点不断构造合法音节节点,进而生成状态空间,遍历算法遍历状态空间可获得所有的切分可能,而当用户进行删除操作时,只需删除部分相关节点。整个状态空间随用户的操作进行局部调整,分布均匀。该算法有利于存在歧义切分问题的整句拼音转换,可从保留下来的所有切分可能中选出一个全局最优的语句候选,保证整句转换的正确性。
Syllable segmentation lays the foundation for sentence pinyin conversion. As a result of the particularity of pinyin syllable segmentation has different ways. One partition result can be got by least participle algorithm so that correctness of sentence pinyin conversion can not be guaranteed. Presents a syllable segmentation algorithm which is composed of two kinds of nodes. One is phoneme node, the other is syllable node. Phoneme nodes are integrated into syllable node. They link with chain to form state space. Can get all the possible partition results by traversal algorithm. When encountering user＇ s deletion operation it only need to delete some nodes related. State space is adjusted locally and equably along with user＇s operation. This algorithm is in favor of sentence conversion with different partition ways. It can get an optimal answer from all the possible partition.
出处
《计算机技术与发展》
2008年第8期35-38,共4页
Computer Technology and Development
基金
安徽大学人才队伍建设经费资助项目(02203105)
关键词
音节切分
切分算法
切分歧义
整句输入
状态空间
syllable segmentation
partition algorithm
various segmentation
sentence input method
state space
