Alt-Texify: A Pipeline to Generate Alt-text from SVG Visualizations

Aspen Belle, Vanessa Goh, Akshay Kumar, Richard Pranjatno, Pui Yip, Umayangani Wickramaratne, Humphrey Obie

2022

Abstract

Data visualizations are used everywhere on the web to convey data and insight. However, interpreting these visualizations is reliant on sight, which poses a problem for the visually impaired who rely on screen readers. Alternative text descriptions are often missing from visualizations or not of a helpful quality, which the screen readers rely on to interpret them for the user. In this short paper, we propose Alt-Texify, a pipeline to generate alternative text descriptions for SVG-based bar, line and scatter charts. Our pipeline classifies the chart type and extracts the data and labels from the SVG code and inserts the relevant information into a description template. Our approach extracts the data and labels deterministically, allowing for factually accurate descriptions 99.74% of the time.

Download


Paper Citation


in Harvard Style

Belle A., Goh V., Kumar A., Pranjatno R., Yip P., Wickramaratne U. and Obie H. (2022). Alt-Texify: A Pipeline to Generate Alt-text from SVG Visualizations. In Proceedings of the 17th International Conference on Evaluation of Novel Approaches to Software Engineering - Volume 1: ENASE, ISBN 978-989-758-568-5, pages 275-281. DOI: 10.5220/0010994600003176


in Bibtex Style

@conference{enase22,
author={Aspen Belle and Vanessa Goh and Akshay Kumar and Richard Pranjatno and Pui Yip and Umayangani Wickramaratne and Humphrey Obie},
title={Alt-Texify: A Pipeline to Generate Alt-text from SVG Visualizations},
booktitle={Proceedings of the 17th International Conference on Evaluation of Novel Approaches to Software Engineering - Volume 1: ENASE,},
year={2022},
pages={275-281},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010994600003176},
isbn={978-989-758-568-5},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 17th International Conference on Evaluation of Novel Approaches to Software Engineering - Volume 1: ENASE,
TI - Alt-Texify: A Pipeline to Generate Alt-text from SVG Visualizations
SN - 978-989-758-568-5
AU - Belle A.
AU - Goh V.
AU - Kumar A.
AU - Pranjatno R.
AU - Yip P.
AU - Wickramaratne U.
AU - Obie H.
PY - 2022
SP - 275
EP - 281
DO - 10.5220/0010994600003176