| 摘要 |
<P>PROBLEM TO BE SOLVED: To generate an object language corresponding word of an unregistered word or phrase which is not registered in a translation dictionary. <P>SOLUTION: The dictionary registration device is provided with: an unregistered word and phrase detection unit 106 which detects an object word or phrase appearing at a frequency of appearance higher than a specified value from a first language corpus and records it together with its appearance position information, and then classifies the object word or phrase as a registered word or phrase or unregistered word or phrase according to a first dictionary from a first language to a second language; a selection unit 114 which selects a registered word or phrase appearing in the first language corpus at a frequency of appearance higher than a specified value within a specified distance range from the unregistered word or phrase; a corresponding word candidate generation unit which generates a corresponding word candidate to possibly be a second language corresponding word of the unregistered word or phrase; a corresponding word retrieval unit which generates a partial list by extracting a plurality of parts including the second language corresponding word of the selected registered word or phrase from a second language corpus; and a corresponding word estimation unit which calculates the degree of probability that the corresponding word candidate is the second language corresponding word of the unregistered word or phrase based upon the degree of proximity between the corresponding word candidate and the second language corresponding word of the registered word or phrase in the partial list, and determines the second language corresponding word. <P>COPYRIGHT: (C)2007,JPO&INPIT |
