Periodic Reporting for period 1 - iDEM (Innovative and Inclusive Democratic Spaces for Deliberation and Participation)
Berichtszeitraum: 2024-01-01 bis 2025-03-31
Deliberative and participatory processes currently lack full legitimacy due to the exclusion and marginalisation of different vulnerable communities from democratic spaces.
Persons with language comprehension difficulties, intellectual or psychosocial disabilities are currently excluded from deliberative and participatory political processes, in civic platforms and the Web, due to the complex language used by policy makers and institutions.
The overall objective of iDEM is to address this major gap in the context of marginalisation and exclusion of people who need support to fully be able to read, write and comprehend complex democratic text in order to be able to fully participate in democratic spaces.
The objectives established in the project are the following.
1. Understand the limitations of current intersectional deliberative and participatory democratic practices in terms of language comprehension and production, and propose accessible and innovative solutions to remove inequalities as to make democratic processes widely accessible
2. Design a solution for making deliberative and participatory processes, at different levels of government, more inclusive, specially for marginalised and vulnerable communities
3. Implement and evaluate state of the art natural language processing technology in the areas of text accessibility and argumentative discourse generation for deliberative processes to facilitate engagement and enhance participation of otherwise marginalised communities
4. Pilot the solution with real users to evaluate the level of inclusiveness and accessibility of the novel proposed democratic spaces
Persons with language comprehension difficulties, intellectual or psychosocial disabilities are currently excluded from deliberative and participatory political processes, in civic platforms and the Web, due to the complex language used by policy makers and institutions.
The overall objective of iDEM is to address this major gap in the context of marginalisation and exclusion of people who need support to fully be able to read, write and comprehend complex democratic text in order to be able to fully participate in democratic spaces.
The objectives established in the project are the following.
1. Understand the limitations of current intersectional deliberative and participatory democratic practices in terms of language comprehension and production, and propose accessible and innovative solutions to remove inequalities as to make democratic processes widely accessible
2. Design a solution for making deliberative and participatory processes, at different levels of government, more inclusive, specially for marginalised and vulnerable communities
3. Implement and evaluate state of the art natural language processing technology in the areas of text accessibility and argumentative discourse generation for deliberative processes to facilitate engagement and enhance participation of otherwise marginalised communities
4. Pilot the solution with real users to evaluate the level of inclusiveness and accessibility of the novel proposed democratic spaces
WP1 undertook a critical analysis of current intersectional deliberative and participatory democratic practices to understand their limitations regarding language comprehension and production, with the goal of proposing accessible solutions to remove inequalities. We examined and analyzed fourteen accessibility-related barriers experienced by people with language communication barriers, focusing on understanding social elements, including gender and ethnic group, that condition the deliberative experience, building an intersectional approach. Moreover, we developed the main theoretical framework based on a threefold analysis: in-depth argumentation on scientific evidence for political equality, an academic discussion on normative democratic theory (including legitimate and epistemological standards), and a study of practical accessibility barriers. These analyses used mixed methods and benefited from discussions at two international, interdisciplinary workshops. The main achievemts are: The identification of barriers analyzed in relation to the stages in which they may appear: before (e.g. recruiting), during (e.g. discussion), and after (e.g. reporting) the deliberative process. The intersectional understanding of how discrimination in deliberation and participation, based on the intersection of factors like ability/disability, gender, race, income, and migration status, acts as an additional barrier. General recommendations for piloting in iDEM aiming at (a) maximizing representativeness through diversity-aware organizations and accessible communication, (b) maximizing deliberative quality through well-designed participatory protocols, accessible information, catering for health and physical needs, and mediation, amd (c) providing guidance for conducting observational studies during piloting to assess the contribution of proposed solutions to deliberation parity and barrier identification. The development of a theoretical Framework provides an informed and analytical framework that will guide the development of the iDEM Use Cases by transferring knowledge into the ideation, design, implementation, and evaluation of the proposed pilots.
In the field of language technology, in WP2 we developed a novel annotation scheme for linguistic difficulties and simplification strategies, specifically for the Catalan language. This scheme integrates expert linguistic knowledge with existing standards. We produced guidelines for the creation and assessment of resources for text simplification and initated the creation of a high-quality, domain-specific multilingual corpus that includes texts in Catalan, Spanish, and Italian, annotated with gold-standard expert simplifications. A comprehensive evaluation framework, incorporating both intrinsic and extrinsic evaluation methodologies, has been established.
As for the technological developments, in WP3 we developed an annotation schema of nine categories based on translation studies to account for transformations observed between original sentences and their easy-to-read adaptations. Transformer-based models were evaluated for multi-class classification, and the developed system provides indications of input elements contributing to sentence complexity. Also implemented the first multilingual lexical simplification system for iDEM, covering Catalan, Italian, and Spanish, in addition to English, using a unified model. This also involved creating the first dataset for lexical simplification in Catalan and a new dataset for Spanish. This was complemented by the developmemt of a trainable easy to read sentence segmentation System. Some experiments were conducted on knowledge distillation techniques, where a large LLM (teacher) is used to mine lexical simplification examples for fine-tuning a small LLM (student). Work is underway to study, test, and analyze several LLMs for text generation to assess their suitability for prompt-based generation.
Research was conducted on how to enhance inclusivity across different stages of a participatory process. To gather information about inclusive moderation and facilitation, data from expert interviews were used to compile information on tools, techniques, formats, adaptation processes, and technology for including vulnerable citizens in iDEM spaces.Moreover we also investigated methods for engaging individuals and communities often excluded from democratic processes due to challenges like language barriers, low literacy, economic problems, discrimination, or limited digital access. Several deliberative and participatory processes that could be adopted by Use Cases in iDEM were researched and planning for the three deliberative use cases was conducted, including process implementation, participant selection and training, meeting structure, time management, topic choice, expert participation, moderation, trust building, iDEM technology use, and reporting. We provided recommendations on how to include hard-to-reach groups in deliberative and participatory processes including preparatory, implementation, and follow-up stages. Factors analysed included co-designing, trust building, facilitation, incentives, physical accessibility, accessible easy-to-understand formats, and post-participation follow-up. A series of recommendations on how to make deliberative and participatory processes more inclusive were produced. They identified measurable objectives for the deliberative process (e.g. quality of deliberation, inclusivity, accessibility) and methods for data collection to assess these objectives (e.g. surveys, interviews). Use Case 1 was fully designed and the evaluation framework was agreed.
In the field of language technology, in WP2 we developed a novel annotation scheme for linguistic difficulties and simplification strategies, specifically for the Catalan language. This scheme integrates expert linguistic knowledge with existing standards. We produced guidelines for the creation and assessment of resources for text simplification and initated the creation of a high-quality, domain-specific multilingual corpus that includes texts in Catalan, Spanish, and Italian, annotated with gold-standard expert simplifications. A comprehensive evaluation framework, incorporating both intrinsic and extrinsic evaluation methodologies, has been established.
As for the technological developments, in WP3 we developed an annotation schema of nine categories based on translation studies to account for transformations observed between original sentences and their easy-to-read adaptations. Transformer-based models were evaluated for multi-class classification, and the developed system provides indications of input elements contributing to sentence complexity. Also implemented the first multilingual lexical simplification system for iDEM, covering Catalan, Italian, and Spanish, in addition to English, using a unified model. This also involved creating the first dataset for lexical simplification in Catalan and a new dataset for Spanish. This was complemented by the developmemt of a trainable easy to read sentence segmentation System. Some experiments were conducted on knowledge distillation techniques, where a large LLM (teacher) is used to mine lexical simplification examples for fine-tuning a small LLM (student). Work is underway to study, test, and analyze several LLMs for text generation to assess their suitability for prompt-based generation.
Research was conducted on how to enhance inclusivity across different stages of a participatory process. To gather information about inclusive moderation and facilitation, data from expert interviews were used to compile information on tools, techniques, formats, adaptation processes, and technology for including vulnerable citizens in iDEM spaces.Moreover we also investigated methods for engaging individuals and communities often excluded from democratic processes due to challenges like language barriers, low literacy, economic problems, discrimination, or limited digital access. Several deliberative and participatory processes that could be adopted by Use Cases in iDEM were researched and planning for the three deliberative use cases was conducted, including process implementation, participant selection and training, meeting structure, time management, topic choice, expert participation, moderation, trust building, iDEM technology use, and reporting. We provided recommendations on how to include hard-to-reach groups in deliberative and participatory processes including preparatory, implementation, and follow-up stages. Factors analysed included co-designing, trust building, facilitation, incentives, physical accessibility, accessible easy-to-understand formats, and post-participation follow-up. A series of recommendations on how to make deliberative and participatory processes more inclusive were produced. They identified measurable objectives for the deliberative process (e.g. quality of deliberation, inclusivity, accessibility) and methods for data collection to assess these objectives (e.g. surveys, interviews). Use Case 1 was fully designed and the evaluation framework was agreed.