本文描述了把扩展句节数最小法进行形态素解析的结果登录在网络[定向图]上,并把这个网络[定向图]保留在计算机的内存中直到修改操作结束,实现了在不同句节切分的各种选择候补中取出所需候补。再根据所选候补去检索网络[定向图]上的路径,就可以得到符合操作者意图的全句的转换结果,从而使日文汉字输入时所需的假名汉字转换操作简单易行,提高了操作性能。

Japanese sentences were recently got by transforming‘kana sequence’into the array of the several syllables. We investigated conversion-algorithm and user interface , having following characteristics. We got other syllable-boundaries without changing-action about those. We adopted fundamentally a method of‘minimizing a sum of syllables in a sentence’. In this method , the result of searching dictionary and of grammatical checking about the chain of the morphemes are maintained in the computer′s memory , so we can easily pick out the candidate of other syllable boundary. We realized to save user′s operating time and got easy user interface.