Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BERTScoreVisualizer: A Web Tool for Understanding Simplified Text Evaluation with BERTScore

Sebastian Jaskowski, Sahasra Chava, Agam Shah

TL;DR

BERTScoreVisualizer is a web application that goes beyond reporting precision, recall, and F1 score and provides a visualization of the matching between tokens and believes that it can help improve the analysis of text simplification systems by specifically showing where generated, simplified text deviates from reference text.

Abstract

The BERTScore metric is commonly used to evaluate automatic text simplification systems. However, current implementations of the metric fail to provide complete visibility into all information the metric can produce. Notably, the specific token matchings can be incredibly useful in generating clause-level insight into the quality of simplified text. We address this by introducing BERTScoreVisualizer, a web application that goes beyond reporting precision, recall, and F1 score and provides a visualization of the matching between tokens. We believe that our software can help improve the analysis of text simplification systems by specifically showing where generated, simplified text deviates from reference text. We host our code and demo on GitHub.

BERTScoreVisualizer: A Web Tool for Understanding Simplified Text Evaluation with BERTScore

TL;DR

BERTScoreVisualizer is a web application that goes beyond reporting precision, recall, and F1 score and provides a visualization of the matching between tokens and believes that it can help improve the analysis of text simplification systems by specifically showing where generated, simplified text deviates from reference text.

Abstract

The BERTScore metric is commonly used to evaluate automatic text simplification systems. However, current implementations of the metric fail to provide complete visibility into all information the metric can produce. Notably, the specific token matchings can be incredibly useful in generating clause-level insight into the quality of simplified text. We address this by introducing BERTScoreVisualizer, a web application that goes beyond reporting precision, recall, and F1 score and provides a visualization of the matching between tokens. We believe that our software can help improve the analysis of text simplification systems by specifically showing where generated, simplified text deviates from reference text. We host our code and demo on GitHub.
Paper Structure (7 sections, 3 equations, 1 figure)

This paper contains 7 sections, 3 equations, 1 figure.

Table of Contents

  1. Introduction
  2. Background
  3. Evaluation of Text Simplification
  4. BERTScore
  5. System Design Goal
  6. Notable Features
  7. Implementation

Figures (1)

  • Figure 1: A screenshot of BERTScoreVisualizer, displaying the matching graph between two text sequences that have similar meanings, but are not exactly the same.