Birds of a Feather and Affinity Group Events

Birds of a Feather and Affinity Group Events Schedule

ACL 2025 has a full lineup of exciting Birds of a Feather (BoF) and affinity group events, which bring together participants around shared research topics, professional experiences, and community affiliations. The hosts of these events look forward to welcoming you to the conference! For any questions about the BoF/affinity group event program, please email acl2025diversity@googlegroups.com.

Mon, Jul 28
Time	Location	Session Title
11:00 - 12:30	1.31-1.32	SomosNLP: The Iberoamerican NLP Community
12:30 - 14:00	1.33	Queer in AI Meet-Up
14:00 - 15:30	1.31-1.32	Mentorship on NLP Research

Tue, Jul 29
Time	Location	Session Title
10:30 - 12:00	1.14	Navigating Challenges in Building Industrial LLM Applications
10:30 - 12:00	1.31-1.32	Humanists in NLP
12:00 - 13:30	1.33	Teaching NLP
14:00 - 15:30	1.14	NLP x Graphs: Where Structure Meets Language
14:00 - 15:30	1.31-1.32	Southeast Asian NLP Community, Projects, and Beyond
14:00 - 15:30	1.33	EquiCL Welcome Session
16:00 - 17:30	1.14	Learning and Reasoning for Structured Data
16:00 - 17:30	1.31-1.32	Multilingualism: from data crawling to evaluation
16:00 - 17:30	1.33	Participatory Design for NLP
16:00 - 17:30	Online (Underline)	Bridging Human Study and LLM Agents for Social Simulation

Wed, Jul 30
Time	Location	Session Title
9:00 - 10:30	1.14	Activations & Embeddings: Cognitive-Neuroscience Methods for LLMs
11:00 - 12:30	1.31-1.32	Mothering the Future — In Life and in AI: Challenges, Support, and the Path Forward for Mothers in Computing
11:00 - 12:30	1.33	Language Technology for Crisis Preparedness and Response (LT4CPR)
12:30 - 14:00	1.14	Ethical Considerations for NLP and CL
12:45 - 14:15	1.31-1.32	Muslims in Machine Learning (MusIML)

Session Descriptions

SomosNLP: The Iberoamerican NLP Community
María Grandury, Selene Báez, Diana Galván, Helena Gómez, Danae Sánchez

In this session, we’ll introduce SomosNLP, the community dedicated to creating and sharing resources that enable and accelerate the development of NLP in Iberoamerican languages. In everyday language, Iberoamerica refers to the countries in the Americas where Spanish or Portuguese is spoken, and also includes Spain and Portugal. Our aim is to connect with individuals from the region, collaboratively exploring current challenges in the field. We’re creating a dynamic space for sharing knowledge, learning from diverse experiences, and devising innovative solutions together. We would like you to support us in our mission! Speakers of less represented languages are especially welcome to join and contribute to this vibrant, multilingual dialogue. We can’t wait to meet you! Find updates and planned activities at https://somosnlp.org/conferencias/acl-2025

Queer in AI Meet-Up
Sabine Weber

This event offers a space for LGBTQIA+ people and their allies to network in a safe and welcoming environment. There will be an introductory talk about the initiative “Queer in AI” and possibly one invited talk about queer-specific issues or related research. After that we will break up into thematic discussion groups. There will be snacks and non-alcoholic drinks.

Mentorship on NLP Research
Oana Ignat, Weijia Shi, Ziqiao Ma

The goal of this session is to provide guidance, share experiences, and facilitate discussions for junior researchers. We believe this session will be a valuable addition to the conference program, contributing to the growth and development of the NLP community while aligning with the conference’s commitment to diversity and inclusion.

Navigating Challenges in Building Industrial LLM Applications
Gauri Kholkar, Aakash Bist, Ratinder Ahuja

Large Language Models (LLMs) are rapidly moving from research labs into real-world industry applications. However, translating cutting-edge models into robust, reliable, and valuable products presents a unique set of practical challenges often distinct from academic pursuits. Issues around data scarcity and quality, model evaluation in production, deployment complexities, scalability, cost management, ethical considerations, ensuring reliability, and seamless integration with existing systems are common hurdles.

This Birds of a Feather session invites researchers, engineers, product managers, and anyone involved in or interested in the practical application of LLMs in industry. Join us for an informal discussion to share experiences, “war stories,” and insights gained from deploying LLMs in the wild. We aim to foster a collaborative environment to discuss common pain points, exchange effective strategies, and collectively brainstorm solutions for navigating the complex landscape of industrial LLM development.

Humanists in NLP
Patrick Sui

This affinity group is for NLP and AI researchers with a prior background or interest in the humanities. As NLP becomes increasingly interdisciplinary - drawing from theories and methods from outside of computer science - it is important to engage with what the humanities (like the social sciences) have to offer in terms of theories and methods to study language, culture, narrative, subjectivity, mind, and emotion. Disciplines like literary and cultural studies, media studies, philosophy, and history have a great deal to offer in terms of studying, building and improving language systems. Topics of interest include (but are not limited to): co intelligence and co-creative systems, narrative understanding, cultural analytics, literary NLP, AI literacy, AI ethics, culture and cognition, etc.

Teaching NLP
Margot Mieskes, Laura Biester, György Kovacs

Conferences are a key component of bringing together researchers to discuss new results and best practices of the NLP field, enabling its progress. Another key component, which is perhaps overlooked, is how we teach future researchers and practitioners. This component is particularly important as our field is rapidly evolving, forcing instructors to constantly reconsider what to include in their curricula (from fundamentals in linguistics to prompting) and how to best use the limited time available in a semester. This topic is highly relevant to many NLP researchers attending NAACL, who have teaching responsibilities as well as research responsibilities. At this informal meetup, we’ll discuss challenges and opportunities in teaching natural language processing in 2025. We hope that this meetup will bring together teachers from different institutions and departments to discuss teaching NLP in varied settings. We invite anyone interested in teaching NLP, from experienced faculty to students who would like to teach NLP in the future.

NLP x Graphs: Where Structure Meets Language
Yuqicheng Zhu, Moritz Plenz

Graphs provide a natural and expressive way to model structure in language—from syntax and semantics to knowledge and discourse. As NLP systems grow in scale and complexity, structured representations are becoming increasingly important for enabling robust reasoning, interpretability, and data efficiency. With the rise of tasks like knowledge-grounded generation, retrieval-augmented language modeling, and graph-structured prompting, graph-based methods are becoming more and more relevant across the NLP landscape.

This BoF session brings together NLP researchers working on or interested in graph-based methods and structured representations. Topics include (but are not limited to) knowledge graphs, semantic and syntactic graphs, AMR, graph neural networks, graph-based retrieval-augmented generation (RAG), structured reasoning with LLMs, and applications of graph theory to NLP. The goal is to foster networking, interdisciplinary exchange, and discussions around open challenges and emerging directions in graph-based NLP. This informal gathering provides a space to meet fellow researchers, share insights from ongoing projects, and identify opportunities for collaboration within the NLP and broader AI community.

Southeast Asian NLP Community, Projects, and Beyond
Fajri Koto, Jan Christian Blaise Cruz, Holy Lovenia, Samuel Cahyawijaya, Alham Fikri Aji, Peerat Limkonchotiwat, M. Reza Qorib

Bringing together Southeast Asian NLP practitioners to explore a collaborative roadmap for advancing NLP research and applications for the SEA region.

EquiCL Welcome Session
Zeerak Talat, Christine de Kock, Fatima Elsafoury, Jackie Lo

The Broad Interest Group on Equity in ACL (EquiCL) is an umbrella organisation for equity and inclusion concerns across the ACL. This session will seek to introduce EquiCL, discuss current and ongoing concerns relating to diversity, and to build a community of people engaged in diversity related initiatives and people interested in supporting such efforts.

Learning and Reasoning for Structured Data
Vivek Gupta, Dan Roth

The BOF session on “Learning and Reasoning for Structured Data” explores how machine learning can better understand and reason over both traditional structured formats—like tables and knowledge graphs—and multimodal structures such as charts, maps, and flowcharts. These data types are central to domains like healthcare, finance, and scientific research, where complex relationships and contextual reasoning are essential. The session highlights new approaches that enhance model accuracy, interpretability, and robustness by aligning learning with relevant evidence rather than superficial patterns. This work is especially beneficial for researchers, data scientists, and AI practitioners seeking to build trustworthy systems capable of navigating and making sense of real-world structured and visual data.

Multilingualism: from data crawling to evaluation
Pinzhen Chen, Andrey Kutuzov, Letiția Pârcălăbescu

The session will cover the complete process of multilingual language modelling, spanning from web data crawling and processing to the development of fairer architectures and trustworthy evaluation protocols.

We will begin with a brief introduction to the HPLT project, providing an opportunity to learn about web-scale multilingual data provision. Participants will be encouraged to raise questions, identify gaps, and suggest improvements for efforts of this nature.

Next, we plan to discuss potential pitfalls in multilingual evaluation, including standards for performance reporting and the use of machine translation in resource creation. Beyond reliance on translation, we will brainstorm scalable approaches for multilingual and multi-cultural evaluation.

Finally, with the goal of achieving fairer language modelling, we will review recent progress in multilingual text synthesis, alongside advancements in architecture and tokenization design that better handle multiple languages.

Participatory Design for NLP
Gavin Abercrombie, Tommaso Caselli

Do you have experience of conducting participatory design for NLP that you’d like to share? Do you want to find out about participatory design for NLP, what it looks like, and how it works? Do you want to get involved in participatory design NLP research, but don’t know where to start?

This session is a chance to discuss the challenges, findings, and successes of engaging in participatory design, and centering the needs of stakeholders in NLP research, and to encourage researchers to engage in participatory design for NLP.

Bridging Human Study and LLM Agents for Social Simulation
Xuan Wang

Recent advances in large language models (LLMs) have opened up new opportunities for simulating complex human behaviors and interactions at scale. These capabilities offer exciting potential for conducting large-scale social simulations, enabling researchers to explore hypotheses in social science, psychology, political science, and more. However, critical gaps remain between traditional empirical human studies and LLM-driven agent simulations. This BoF aims to bring together researchers across NLP, social sciences, HCI, and AI ethics to discuss key questions: How can we ensure that LLM agents faithfully represent human behaviors? What are the methodological, ethical, and technical challenges in integrating human data and LLM simulations? How can social simulations with LLMs complement or extend human subject research?

Activations & Embeddings: Cognitive-Neuroscience Methods for LLMs
Giovanni Franco Gabriel Marraffini

Groundbreaking research reveals that transformer features predict human brain activity with remarkable precision. fMRI, MEG, and EEG studies show layer-by-layer correspondences between LLMs and cortical language networks, with larger models yielding stronger neural alignment. Meanwhile, mechanistic interpretability is uncovering discrete “circuits” within LLMs, offering unprecedented insights into language processing.

The Challenge: These advances span three disconnected communities—NLP researchers, cognitive scientists, and neuroscientists. This BoF unites these communities to accelerate progress toward brain-informed language technology.

What You’ll Gain: Learn from peers about cutting-edge protocols for neural encoding/decoding, RSA analysis, behavioral alignment tasks, and interpretability techniques. Discover which transformer components (attention heads, MLPs, memory blocks) correspond to specific brain regions and processing timelines. Access curated datasets (Narratives fMRI, Cross-Subject EEG), evaluation frameworks (BrainScore, NeuroBench), and open-source tools. Form cross-disciplinary collaborations for future workshops, shared tasks, and research projects. Explore how neuroscience-informed probes improve model alignment and bias detection.

Target Audience: NLP researchers interested in cognitive plausibility; neuroscientists and psycholinguists working with language models; ML practitioners developing brain-inspired architectures; early-career researchers bridging computational and cognitive sciences; anyone curious about AI-human cognition intersections.

Interactive Session (90 minutes): Lightning Rounds (10 min) feature quick “What I study/What I need” pitches from attendees. Breakout Tables (35 min) include thematic small-group discussions on neuroimaging & LLMs, behavioral alignment methods, mechanistic interpretability, and cross-linguistic brain studies. Synthesis & Planning (45 min) covers collaborative research roadmapping, concrete next steps for dataset sharing and collaboration, and community building.

Expected Outcomes: Living resource hub (GitHub repository) with datasets, analysis pipelines, and expert contacts. Cross-disciplinary mailing list for ongoing collaboration. Seed teams for ACL 2026 workshop on “Brain-Aligned LLM Evaluation.” Enhanced networking for researchers at the NLP-neuroscience intersection.

Why ACL 2025: ACL uniquely brings together leading LLM researchers alongside cognitive scientists and neuro-NLP experts. Vienna’s central location makes this the perfect venue to catalyze international collaboration and chart the future of brain-informed language technology.

Mothering the Future — In Life and in AI: Challenges, Support, and the Path Forward for Mothers in Computing
Narjis Asad

The pursuit of research, especially in the fast-paced world of AI and Machine Learning, should not come at the cost of parenthood. Yet, for early-career female researchers, these two paths often feel mutually exclusive. I am a PhD researcher in AI and the mother of a young child. This dual identity, though deeply fulfilling, remains exceedingly rare in our field. Why? Because many women delay motherhood until after their PhDs — often at personal and biological cost — or leave their programs altogether post-pregnancy. Others choose (or feel compelled to choose) alternatives like adoption or pet parenting simply because they don’t feel secure or supported enough to start a family early in their academic careers.

The reasons are, sadly, genuine and pressing: the relentless demands of research, the fear of losing momentum during maternity leave, financial precarity, and the silent but persistent doubt — Can I truly do justice to both roles?

In speaking with fellow researchers, I’ve realized I’m not alone in navigating this tension. Yet few spaces exist where these stories are validated or these challenges openly addressed. While our community rightly celebrates women’s achievements in computing, it often overlooks the structural and emotional costs behind them, especially for those who are, or wish to become, mothers.

Sacrifices are part of any high-achievement path, yes. But no woman should have to sacrifice her biological clock or risk her future child’s health in order to become an accomplished researcher in computing. That should never be the price of success or excellence in research.

This 90-minute hybrid affinity session will open up a much-needed conversation: What structural changes and community support do we need to make motherhood a supported and celebrated norm for early-career researchers in computation?

We invite mothers in computer science, early-career, and established to share their stories, offer solidarity, and envision a more inclusive future. I’ve experienced firsthand the challenges of combining early motherhood with an intense research career, and I’m deeply passionate about improving support systems for “Mothers in Computation” in India and globally.

Whether you are a mother, a future parent, an ally, or a mentor, this session is for you. Join us as we imagine and begin to build a world where embracing motherhood during a research career is not only possible but also joyous and empowering.

Language Technology for Crisis Preparedness and Response (LT4CPR)
Belu Ticona, Antonios Anastasopoulos, Will Lewis, Fei Xia.

How can we use language technology to prepare for the next crisis? In a world with more natural disasters and socio-political conflicts, crisis communication is crucial to be timely, accurate and trustworthy. Governmental institutions, NGOs and affected populations communicate with each other through formal and informal channels, involving not only different languages, cultures and modalities, but also different linguistic registers, signal quality and response time. How can we develop language technologies that facilitate the management of crises across local and national borders? How do the communication needs of stakeholders evolve across each stage of the crisis management cycle? In this BoF, we will discuss open questions and challenges that make crisis preparedness and response an emergent field. We invite anyone interested in low-resource languages, machine translation, multicultural NLP, climate change, crisis management and language in general!

Ethical Considerations for NLP and CL
Margot Mieskes, Karën Fort, Fanny Ducel, Clémentine Bleuze, Aurélie Névéol

Ethical considerations touch a wide range of research and researchers in Natural Language Processing and Computational Linguistics. In this BoF session we would like to bring together a wide variety of people who are working in these topics or are interested to learn more about them. One goal is to grow our list of available material in the Ethics in NLP reading list here: https://ethics.aclweb.org/resources/ethics-reading-list/ Another is to raise interest in this topic, grow awareness and maybe also recruit more people willing to support ethics reviewing.

Muslims in Machine Learning (MusIML)
Ehsaneddin Asgari, Suleiman Ali Khan, Ahmed Youssef

MusIML is an independent grassroots initiative committed to advancing equity, representation, and collaboration for Muslims in AI/ML. The organization has hosted three successful workshops at NeurIPS (2020, 2023, 2024), with the upcoming 4th edition co-located with ICML 2025. It also organizes social and mentoring events at major conferences including ICLR and involvement in AISTATS. Our initiative has a few sister programs including the Fatima Fellowship, a global mentorship program, with a vibrant Slack community of 200+ members. More details can be found at www.musiml.org.