Explainable Point-Based Document Visualizations

TL;DR

This paper explores using keyword extraction methods, especially YAKE!, to label clusters in point-based document visualizations like t-SNE and UMAP, enhancing interpretability of data maps.

Contribution

It introduces a novel approach of applying keyword extraction to label document clusters in visualizations, comparing multiple methods and highlighting YAKE!'s effectiveness.

Findings

YAKE! outperformed other keyword extraction methods

TF-IDF was more effective than graph and embedding-based techniques

Keyword labeling improves interpretability of data maps

Abstract

Two-dimensional data maps can visually reveal information about the relations between data instances. Popular techniques to construct data maps are t-SNE and UMAP. The resulting point-based visualizations, though, provide information only through their interpretation. We here consider a set of abstracts from the articles on longevity to argue for using keyword extraction methods to label clusters of documents in the map. Among the considered approaches, the best results were obtained by recently proposed YAKE!. Surprisingly, a classical TF-IDF term ranking outperformed graph and embedding-based techniques.