Participation of Lourdes Moreno in the "Artificial Intelligence and Accessibility Research Symposium" of the W3C, in the panel "Natural Language Processing for Accessible Communication"
On January 10 and 11, the "Artificial Intelligence (AI) and Accessibility Research Symposium" was held, within the framework of the W3C WAI initiative with the project "Communities of Practice (WAI-CooP) the W3C Accessible Platform Architectures (APA) Working Group”.
This symposium brought together researchers to identify current challenges and opportunities posed by the increasing use of AI in relation to accessibility and to explore how ongoing research can leverage and improve accessibility.
There were different panels related to AI and Natural Language Processing (NLP) technology applied to accessibility challenges such as computer vision for accessibility in media, PLN for accessibility in media, machine learning for the evaluation of web accessibility, and PLN for accessible communication. In addition, there were talks by the great accessibility experts Jutta Treviranus and Shari Trewin.
Lourdes Moreno participated in the PLN and communication panel, introducing the most widely used language models today. In addition, he spoke about the use of PLN methods to improve accessibility in communication through the task of simplifying texts to clear language and easy reading, as well as the task of generating summaries as an accessibility resource to be able to access relevant information from long and tedious texts. She identified challenges such as avoiding and correcting for disability bias in the datasets that algorithms learn from in order to avoid disadvantages to people with disabilities. In addition, she spoke about the lack of resources such as corpora and dictionaries in the field of accessibility prepared by accessibility experts and the participation of people with cognitive disabilities. Finally, she pointed out the few language resources in Spanish compared to English. She introduced the EASIER corpus created by the HULAT group within the framework of the EASIER project as an example of good practice. This corpus has been prepared by experts in easy reading and with the participation of people with intellectual disabilities and the elderly.
Organizers will publish a report with the conclusions and findings.