| 摘要 |
PROBLEM TO BE SOLVED: To prevent a precision problem of increase in similarity due to importance proximity between search words belonging to different technological fields. SOLUTION: This document search device is provided with an objective document input part 2A receiving data of an input document (objective document), a morphological analysis part 2B performing morphological analysis on the objective document to resolve it into search words, a technological field classification part 2C finding similarity between the objective document and a reference document from a characteristic vector using search word importance degrees, which are decided based on appearance circumstances of the respective search words in the objective document and the reference document and number of search word appearance documents in a population, as vector components for deciding the technological field, to which the objective document belongs, according to the found similarity, and a technological field theme catalogue table 9 assuming a previously defined set of search words relevant to each technological theme in each technological field as a document (a catalogue document) and storing information about a characteristic vector for the catalogue document. In the technological filed classification part 2C, the reference document means the catalogue document. COPYRIGHT: (C)2006,JPO&NCIPI
|
