Emerging Trends in Artificial Intelligence for Digital Accessibility: A Topic Modeling Analysis

Abstract

In past decades, extracting key themes from vast amounts of unstructured data has facilitated the identification of meaningful categories, called topics, from the coding of textual content. While studies have examined the intersection of artificial intelligence and different domains, there is limited research focusing on understanding trends of AI for digital accessibility. This study aims to examine and identify domain areas, themes, and the interconnectivity of AI in accessibility by analyzing 1,479 papers. Using a structured topic modeling approach, the study analyzes emerging themes and topics. A network analysis method with the Fruchterman-Reingold algorithm was also used to understand the interaction and interconnectedness of these topics. The findings revealed 23 relevant topics and their corresponding keywords in the field of AI for digital accessibility, organized into four network communities and the connections between them. The findings highlight an imbalance in research priorities, indicating that current work heavily addresses sensory disabilities (particularly visual impairment), while critical areas such as AI for cognitive disabilities and the ethical challenges of AI in accessibility receive considerably less attention. The human-centered topic cluster were shown to be central to the research network.

Ähnliche Arbeiten