| 摘要 |
PROBLEM TO BE SOLVED: To improve the analysis accuracy by obtaining an optional number of candidates of word strings forming a Japanese sentence in order of higher probability and showing an optional number of candidates of the word strings in order of higher simultaneous probability based on a word division model and a word model and after input of the Japanese sentence. SOLUTION: A Japanese sentence is inputted (S1). An optional number of candidates of word strings forming the Japanese sentence are obtained in order of higher probability based on a word division model which gives the simultaneous probability of word strings forming a sentence based on the word two-set probability and a word model which gives the simultaneous probability of characters forming a word based on the word length probability and the intra- word character two-set probability (S2). Then, an optional number of candidates of word strings forming a sentence are shown in order of higher simultaneous probability of word strings (S3). In such a constitution, the analysis accuracy is improved even when the words which are not registered in a dictionary are included in the input sentences. |
