For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Efficient Discovery of Highly Interrelated Users in One-Way Communications
Jihwan SONG Deokmin HAAM Yoon-Joon LEE Myoung-Ho KIM
IEICE TRANSACTIONS on Information and Systems
Publication Date: 2011/03/01
Online ISSN: 1745-1361
Print ISSN: 0916-8532
Type of Manuscript: LETTER
Category: Artificial Intelligence, Data Mining
data mining, sequential pattern, Apriori property violation,
Full Text: PDF(851KB)>>
In this paper, we introduce a new sequential pattern, the Interactive User Sequence Pattern (IUSP). This pattern is useful for grouping highly interrelated users in one-way communications such as e-mail, SMS, etc., especially when the communications include many spam users. Also, we propose an efficient algorithm for discovering IUSPs from massive one-way communication logs containing only the following information: senders, receivers, and dates and times. Even though there is a difficulty in that our new sequential pattern violates the Apriori property, the proposed algorithm shows excellent processing performance and low storage cost in experiments on a real dataset.