Towards More Accessible Scientific PDFs for People with Visual Impairments: Step-by-Step PDF Remediation to Improve Tag Accuracy

2025-03-28

AI Summary

PDF documents exhibit poor accessibility for visually impaired users, particularly due to the lack of semantic tagging for mathematical formulas and inadequate screen reader support. Method: This paper introduces PAVE 2.0—a novel system enabling non-expert users to perform accessible mathematical formula processing in PDFs. It integrates rule-guided interactive structural analysis, heuristic label recommendation, and a lightweight OCR–formula semantic parsing module to establish a progressive, interpretable PDF tagging guidance mechanism. Contribution/Results: Experiments show that label accuracy improves from 42.0% to 80.1% for experts and from 39.2% to 75.2% for novices. Fifteen of 19 participants expressed willingness to continue using PAVE 2.0, and all recommended its adoption in accessibility training. The core contribution lies in substantially lowering the technical barrier to PDF semantic annotation, achieving an inclusive breakthrough in automated, accessible mathematical content generation.

Technology Category

• Humans and AI: AI for Accessibility
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Computer Vision: Visual Reasoning & Symbolic Representations

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

PDF inaccessibility is an ongoing challenge that hinders individuals with visual impairments from reading and navigating PDFs using screen readers. This paper presents a step-by-step process for both novice and experienced users to create accessible PDF documents, including an approach for creating alternative text for mathematical formulas without expert knowledge. In a study involving nineteen participants, we evaluated our prototype PAVE 2.0 by comparing it against Adobe Acrobat Pro, the existing standard for remediating PDFs. Our study shows that experienced users improved their tagging scores from 42.0% to 80.1%, and novice users from 39.2% to 75.2% with PAVE 2.0. Overall, fifteen participants stated that they would prefer to use PAVE 2.0 in the future, and all participants would recommend it for novice users. Our work demonstrates PAVE 2.0's potential for increasing PDF accessibility for people with visual impairments and highlights remaining challenges.

Problem

• Improving PDF accessibility for visually impaired users
• Providing step-by-step PDF remediation without expert knowledge
• Enhancing tag accuracy for screen reader navigation

Innovation

• Step-by-step PDF remediation process
• Alternative text for mathematical formulas
• PAVE 2.0 prototype evaluation