Linguistic Geometries for Unsupervised Dimensionality Reduction

TL;DR

This paper investigates the use of linguistically informed geometries to improve unsupervised dimensionality reduction and visualization of high-dimensional text data.

Contribution

It introduces a novel approach that incorporates domain knowledge and linguistic resources into the geometric framework for dimensionality reduction.

Findings

Enhanced visualization quality with linguistically informed geometries

Improved preservation of semantic relationships in low-dimensional embeddings

Demonstrated effectiveness across multiple text datasets

Abstract

Text documents are complex high dimensional objects. To effectively visualize such data it is important to reduce its dimensionality and visualize the low dimensional embedding as a 2-D or 3-D scatter plot. In this paper we explore dimensionality reduction methods that draw upon domain knowledge in order to achieve a better low dimensional embedding and visualization of documents. We consider the use of geometries specified manually by an expert, geometries derived automatically from corpus statistics, and geometries computed from linguistic resources.