Semantic analysis in word vector spaces with ICA and feature selection
Tiina Lindh-Knuutila, Jaakko Väyrynen, Timo Honkela; Proceedings of KONVENS 2012 (Main track: oral presentations), pp. 98-107, September 2012.
In this article, we test a word vector space model using direct evaluation methods. We show that independent component analysis is able to automatically produce meaningful components that correspond to semantic category labels. We also study the amount of features needed to represent a category using feature selection and syntactic and semantic category test sets.