This paper explores the possibilities of using independent component analysis (ICA) for features extraction that could be applied to word sense induction. Two different methods for using the features derived by ICA are introduced and results evaluated. Our goal in this paper is to observe whether ICA based feature vectors can be efficiently used for word context encoding and subsequently for clustering. We show that it is possible, further research is, however, necessary to ascertain more reliable results.
|出版狀態||已發佈 - 2007|
|事件||19th Conference on Computational Linguistics and Speech Processing, ROCLING 2007 - Taipei, 臺灣|
持續時間: 2007 9月 6 → 2007 9月 7
|其他||19th Conference on Computational Linguistics and Speech Processing, ROCLING 2007|
|期間||2007/09/06 → 2007/09/07|
ASJC Scopus subject areas