Akademik Bilişim 2011 - İnönü Üniversitesi

Başlık:Use of Growing Self Organizing Maps in Text Mining
Yazar(lar):Yazar #1
Ad Soyad: Zafer İşcan
Kurum: İstanbul Teknik Üniversitesi
Ülke: Turkey
E-Posta: iscanz__at__itu.edu.tr

Anahtar Kelimeler:Self-Organizing Map, Data Mining, Text Mining
Özet:Use of Growing Self Organizing Maps in Text Mining In the study, growing self-organizing map (GSOM) based two papers that perform different approaches for decreasing clustering time in text mining are examined. In the first paper, a clustering method using growing self organizing map in two steps is presented. Most significant amount of the clustering process is divided into sub-processes which can be performed in different computers by using the evolving grid technology. Therefore, a quick analysis of the acquired information becomes possible. Performance of the proposed method is close to the traditional approaches. However, application time is improving 15 times. In the second paper, a new algorithm (HDGSOMr) based on a growing version of SOM is introduced. The new algorithm adds randomness to the self-organizing process in order to generate more qualified clusters in a few iterations by using smaller neighborhood. Thus, total time of the operation decreases significantly.
Konu(lar):Veri Madenciliği
Dosya:106.doc