| 摘要 |
PROBLEM TO BE SOLVED: To detect a mail including many words having a high appearance frequency along with a topic keyword as a mail related to the topic keyword even if the mail does not include the topic keyword. SOLUTION: A mail sorting system comprises: a thread integration means for integrating a group of mails having high possibility of including a common topic as a thread; a sorting word candidate extraction means for extracting a sorting word candidate from the mail text of each thread; a sorting word deriving means for deriving a sorting word effective for sorting mails; a mail sorting feature quantity deriving means for deriving a mail sorting feature quantity indicating similarity among threads, based on whether the sorting word is included in a thread; and a clustering means for performing clustering based on the mail sorting feature quantity and grouping a thread where the same topic is included and a group of mails for extracting as a cluster. COPYRIGHT: (C)2009,JPO&INPIT
|
