Text documents are complex high dimen-sional objects. To effectively visualize such data it is important to reduce its di-mensionality and visualize the low dimen-sional embedding as a 2-D or 3-D scatter plot. In this paper we explore dimension-ality reduction methods that draw upon domain knowledge in order to achieve a better low dimensional embedding and vi-sualization of documents. We consider the use of geometries specified manually by an expert, geometries derived automat-ically from corpus statistics, and geome-tries computed from linguistic resources.
展开▼
