| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a new symbol string dividing technology capable of dividing even a symbol string the dividing method of which is unknown while following a simple configuration. <P>SOLUTION: A symbol element to be a target of coupling is generated by dividing an input symbol string into symbol units, and the appearance frequencies of pairs of the symbol element and adjacent symbol element are then acquired by referring to basic symbol strings. Those appearance frequencies are used to evaluate the degree of coupling of the symbol element pairs, thereby determining the symbol element pairs indicating the high degree of coupling among those symbol element pairs as targets to be coupled. Then, the symbol element pairs indicating the high degree of coupling are coupled to be a new symbol element, and this process is repeated until the symbol element pairs indicating the high degree of coupling are consumed, thereby dividing the input symbol string. Because symbol strings can automatically be divided only by utilizing symbol units appearing in the basic symbol strings and the appearance frequencies of the symbol strings, even the symbol string the dividing method of which is unknown can be divided while following the simple configuration. <P>COPYRIGHT: (C)2004,JPO |
