| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To estimate abbreviation likelihood based on a characteristic of an abbreviation candidate and a division position into element words, without depending on an abbreviation generation rule. <P>SOLUTION: This abbreviation estimation device includes a first acquisition means 101 for acquiring a character sequence serving maybe as the abbreviation candidate, and dividable position information for dividing the character sequence, a calculation means 102 for calculating a plurality of the first features indicating a feature of the character sequence drawn out from the character sequence and the dividable position information, a storage means 106 for storing the second feature belonging to a desired category out of the categories of an index indicating a concept of the feature, and a likelihood corresponding to the second feature, a second acquisition means 103 for comparing the plurality of first features with the second feature, to acquire the likelihood corresponding to the same feature, and a calculation means 104 for calculating abbreviation candidate order that is order indicating rightness of the character sequence as the abbreviation with respect to a formal name, based on the likelihood acquired by the second acquisition means. <P>COPYRIGHT: (C)2010,JPO&INPIT</p> |
