摘要 |
PROBLEM TO BE SOLVED: To allow a publisher to determine whether a known document is a document such as advertisement undesired by a user adaptatively even when it is difficult to be determined from document contents especially. SOLUTION: Documents described by the same publisher are acquired from a learned document storage means, the similarity of two adjacent documents is calculated, a classification rule for classifying a publisher whose classification result is unknown is learned by using average similarity which is the featured value of each publisher and the known classification result of the publisher, and the learned result is stored in a classification rule storage means. Then a document group is acquired from a classification target document storage means in which a document group described by a publisher whose classification result is unknown is stored, each document is analyzed, the featured value of the document is calculated, and while referring to the classification rule storage means, the classification of the publisher is determined. COPYRIGHT: (C)2007,JPO&INPIT
|