The 1st Multidisciplinary Workshop on NLP for Democracies (MultiDems)
March 13, 2027. Co-located with EACL 2027 in Athens, Greece.
(Apologies for cross-posting)
Recent advances in Natural Language Processing (NLP) and Large Language Models (LLMs) create new opportunities for supporting democratic processes, from large-scale citizen consultation and deliberation to debate summarization, preference aggregation, and access to political information. At the same time, their use raises fundamental questions about fairness, diversity, political representation, transparency, and whose voices and perspectives are amplified or marginalized.
The 1st Multidisciplinary Workshop on NLP for Democracies (MultiDems) aims to provide a dedicated forum for research at the intersection of NLP, political science, social science, computational social choice, and related disciplines. We particularly encourage the submission of work that moves beyond predominantly US-centric, monolingual, or two-party settings and studies the linguistic, institutional, and political diversity of democratic systems worldwide.
Topics of Interest
Topics include, but are not limited to:
-
Social choice and computational social choice in NLP
-
Agent-in-the-loop and AI-assisted deliberation
-
Summarization or moderation of political debates, deliberations and citizen consultations
-
Preference inference and aggregation of citizensâ preferences from natural language
-
Fairness and bias in political text classification and NLP systems
-
Multilingual NLP for democratic systems, including multilingual and multi-party contexts
-
Critical evaluation of AI and NLP systems used in democratic processes
-
Representation of diverse, minority, and critical voices in AI systems
-
NLP for participatory and deliberative democracy
-
Political and parliamentary NLP
-
Argument mining and opinion aggregation in democratic settings
-
Evaluation methods and benchmarks for democratic applications of NLP
-
Human-centered and participatory approaches to AI for democracy
-
Ethical, societal, and institutional implications of deploying NLP and LLM systems in democratic contexts
We particularly encourage interdisciplinary work bringing together NLP with political science, sociology, computational social science, computational social choice, HCI, public policy, law, or related fields.
In addition to conventional research papers, we welcome position and conceptual papers, methodological contributions, negative results, datasets and resources, evaluation studies, critical analyses, and interdisciplinary empirical work.
Submission
MultiDems will accept both archival and non-archival submissions.
Archival long papers: up to 8 pages of content, excluding references and appendices.
Archival short papers: up to 4 pages of content, excluding references and appendices.
Archival submissions must describe original work that has not been published elsewhere and, if accepted, will appear in the workshop proceedings. Archival submissions require one author to volunteer as a potential reviewer. Potential reviewers will only be contacted if the existing reviewer pool is not large enough for the number of submissions.
Non-archival extended abstracts: up to 4 pages, excluding references. These submissions provide an opportunity to present and discuss work without publication in the workshop proceedings, including relevant work published or accepted elsewhere where permitted by the original venue.
Both long and short submissions may contain technical, empirical, theoretical, conceptual, or critical contributions.
All submissions must use the official ACL formatting templates. Papers will be submitted through OpenReview and will go through a double-blind review process.
MultiDems will accept:
-
Direct submissions: submit via https://openreview.net/group?id=eacl.org/EACL/2027/Workshop/MultiDems
-
Papers previously reviewed through ACL Rolling Review (ARR): commit your ARR paper to the workshop via https://openreview.net/group?id=eacl.org/EACL/2027/Workshop/MultiDems_ARR_Co...
-
Shared-task system-description papers (see Shared Task Section below): submit via https://openreview.net/group?id=eacl.org/EACL/2027/Workshop/MultiDems_Shared...
For direct submissions, we will seek reviewers across disciplinary backgrounds and, where appropriate, aim to combine NLP/computer-science and social-science perspectives in the review process. Please also note that shared-task system-description papers follow a separate submission timeline, detailed in the Important Dates section below.
Double Submission Policy
Papers submitted to the archival track must not have been previously published and must not be under review at any other venue (conference, workshop, or journal) at the time of submission. Submissions that are found to be under review elsewhere, or that substantially overlap with work published or under review elsewhere, will be desk-rejected.
Generative AI Policy
MultiDems follows the [ACL policy on AI writing assistance](https://www.aclweb.org/adminwiki/index.php/ACL_Policy_on_Publication_Ethics#...) and will require authors to state the usage of AI during submission. Additionally, MultiDems follows a zero tolerance policy on hallucinated references. Submissions will go through an automatic false references checker, and positive cases will be manually reviewed. All papers with at least one confirmed hallucinated reference will be desk rejected.
Shared Task: Summarization of Citizen Consultations
MultiDems will also host an associated shared task on Text Summarization for Online Citizen Consultations. Online consultations enable large-scale democratic participation by collecting citizens' proposals and their votes on public issues (e.g., âWe should incorporate the right to clean air into our constitution.â). Turning these contributions into concise, accessible summaries is essential for informing public deliberation and decision-making. Although large language models offer promising opportunities for automating this process, they also raise important questions: Whose proposals are represented? Are minority viewpoints adequately reflected? Does the summary accurately convey both the meaning of proposals and their level of public support?
Task: Participants will develop systems that produce a single, concise, and coherent textual summary of the entire citizen consultation, based on its proposals and associated voting data.
Data & Evaluation: The task introduces a multilingual benchmark based on data from a multilingual European citizen consultation. The dataset is now publicly available at https://github.com/SalimHFX/Multidems. It contains 393 citizen proposals across 9 languages, together with a completed voting matrix representing the observed and inferred preferences of 934 participants. Beyond conventional summary quality, generated summaries will be evaluated using democratically grounded criteria, including proportional representation of citizensâ preferences, semantic fidelity to the original proposals, and coverage of the opinion space. Importantly, trade-offs between metrics are expected and constitute a central motivation for this shared task: different principles of democratic representation may conflict, and optimizing one criterion may compromise another. The task therefore aims not only to identify effective summarization methods, but also to investigate how alternative approaches navigate these trade-offs. The exact evaluation metrics, evaluation scripts, and baseline implementationwill be released soon. Instructions for submitting generated summaries will also be announced shortly.
Participation: Participation is open to researchers and practitioners from NLP/ML/AI, computational social choice, political science, and related fields. Teams may participate in the competition without submitting a paper. Participants are also encouraged to submit archival system-description papers presenting their approaches, experimental results, and analyses of democratic evaluation criteria. Selected teams, including those performing particularly well across different evaluation dimensions, will be invited to present their work at the workshop.
How to participate in the shared task: Register using this form https://forms.gle/FHe4SGvofhYWAtwZA. Registered participants will be invited to join our Google Group to receive updates and announcements about the shared task.
Important Dates
All deadlines are 11:59 PM Anywhere on Earth (UTCâ12).
-
Direct paper submission deadline: December 23, 2026
-
ARR commitment deadline: January 18, 2027
-
Notification of acceptance: January 22, 2027
-
Camera-ready papers due: February 2, 2027
-
Pre-recorded video due: February 12, 2027
-
MultiDems workshop date: March 13, 2027
Dates specific to the shared task are the following:
-
Data release: November 20, 2026
-
Summaries submission deadline : Late December 2026 (TBD)
-
System-description paper submission deadline: Mid January 2027 (TBD)
-
Notification of acceptance: Late January 2027 (TBD)
Invited Speakers
We are delighted to welcome:
Niclas Boehmer, Assistant Professor, Hasso Plattner Institute, University of Potsdam Researching computational social choice and foundations for fair AI/NLP systems.
Judith Möller, Full Professor, Leibniz Institute for Media Research, University of Hamburg Researching algorithms, recommender systems, political communication, and the effects of AI on democratic dynamics.
Workshop Format
MultiDems is planned as a one-day hybrid workshop, with participation possible both in person at EACL 2027 in Athens and remotely. The program will include invited talks, oral presentations of selected papers, poster presentations, shared-task results, and opportunities for interdisciplinary discussion.
Organizers
François Yvon â Sorbonne UniversitĂ© / CNRS, France Jean-Philippe Cointet â Sciences Po mĂ©dialab, France Benjamin Piwowarski â Sorbonne UniversitĂ© / CNRS, France GaĂ«l Lejeune â Sorbonne UniversitĂ©, France Salim Hafid â Sciences Po mĂ©dialab, France Paul Lerner â Sorbonne UniversitĂ©, France Tanise Ceron â Bocconi University, Italy Eoghan Cunningham â University College Dublin, Ireland Nazanin Shafiabadi â Sorbonne UniversitĂ©, France Pierre-Antoine Lequeu â Sorbonne UniversitĂ©, France Carlo Santagiustina â Inria / Sciences Po mĂ©dialab, France LaurĂšne Cave â Sorbonne UniversitĂ©, France
Workshop website
https://sites.google.com/view/multidems/home
Contact
multidems@googlegroups.com
In case of trouble with the google groups email, you can contact the organizers at salim.hafid@proton.me and lequeu@isir.upmc.fr.