Word sense induction using independent component analysis

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.