Authors: Saggion, O’Flaherty, Blanchet, Sharoff, Sanfilippo, Muñoz, Gollegger, Rascón, Martí, Szasz, Bott, Sayman
Source: SEPLN-CEDI2024: Seminar of the Spanish Society for Natural Language Processing at the 7th Spanish Conference on Informatics
Question: A newly Horizon-funded EU research initiative, IDEM (Innovative and Inclusive Democratic Spaces for Deliberation and Participation) is studying how deliberative and participatory democratic practices can be made more inclusive and accessible, particularly for marginalized and vulnerable communities with limited language skills?
Significance: Millions of people in the EU and globally struggle with language barriers (low levels of literacy, intellectual disabilities, dyslexia, aphasia, temporary impairments, elderly or those affected by a significant cognitive, physical, or sensory decline) that restrict their full participation in democracy. The iDEM project seeks to break these linguistic barriers and create a new era of inclusive and participatory democratic spaces for marginalized communities. By promoting inclusivity and representation, iDEM aims to strengthen the legitimacy of deliberative and participatory processes.
Method: The iDEM project will adopt a user-centered approach to investigate the linguistic barriers hampering participation in deliberative processes. It will create parallel domain-specific annotated datasets for text simplification in Catalan, Spanish, and Italian. Natural language processing models will be fine-tuned to identify and classify sources of text complexity and simplify them accordingly. Tools for generating coherent discourses using fine-tuned Large Language Models will also be developed.
Experiment: The iDEM project will develop and implement three use cases, each focusing on a different language (Catalan, Italian, or Spanish), intersectionality (people with disabilities, migrants, the elderly), and deliberative approach (citizen's assembly, mini-public, consultation). The use cases will follow a detailed prototyping process to collect user and institutional requirements, develop text simplification functionalities, and create tools for engaging hard-to-reach and vulnerable populations. Facilitators of the deliberative and participatory processes will be interviewed to understand the challenges faced by marginalized groups in political participation.
Findings: The iDEM project has just launched, and the findings will be based on the evaluation of the use cases by external experts. The evaluation will assess the inclusiveness and accessibility of the proposed solution. The project aims to create next-generation multilingual models that automatically adapt texts to the needs of the people and provide AI tools for unbiased communication.
Reflection for Democracy: The iDEM project will address how large language models can promote greater inclusivity in participatory processes. It draws attention to the positive potential for using AI to promote inclusion. The study has the potential to create novel participatory spaces that use text simplification technology and natural language generation. By facilitating the participation of marginalized communities in democratic processes, iDEM may contribute to a more inclusive and representative democracy. The project's success could pave the way for a new era of deliberative and participatory practices that prioritize the voices and needs of all citizens, regardless of their language skills or background overcoming outdated conceptions of citizen competence.