NB-REAL Workshop 2025

About the Workshop

This workshop focuses on the responsible evaluation and alignment of Large Language Models (LLMs) for Nordic and Baltic languages.

Develop ethically-grounded, challenging benchmarks in Nordic and Baltic languages.
Create and encourage the development of culturally sensitive alignment datasets.
Promote research and collaboration on responsible LLM alignment techniques.
Discuss sustainable and ethical methods for collecting evaluation and alignment data.
Explore challenges specific to evaluation and alignment of LLMs for Nordic and Baltic languages.

For this workshop, Nordic countries include Åland Islands, Denmark, Faroe Islands, Finland, Greenland, Iceland, Norway and Sweden. Baltic countries include Estonia, Latvia, and Lithuania.

Workshop Organizers

Hafsteinn Einarsson

Associate Professor in Computer Science, University of Iceland

hafsteinne@hi.is

Annika Simonsen

PhD Student, University of Iceland

annika@hi.is

Dan Saattrup Nielsen

Senior AI Specialist, Alexandra Institute

dan.nielsen@alexandra.dk

Program Committee

Barbara Scalvini

University of the Faroe Islands

Elena Callegari

University of Iceland

Haukur Barri Símonarson

Miðeind

Iris Edda Nowenstein

University of Iceland

Kenneth Enevoldsen

Aarhus University

Lars Bungum

Norwegian University of Science and Technology

Steinunn Rut Friðriksdóttir

University of Iceland

Program Design

Half-day workshop on March 2, 2025 (pre-conference). Hybrid event with both on-site and online participation available.

Time	Activity
09:00-09:15	Opening Remarks
09:15-10:00	Keynote Speaker (45 min)
10:00-10:20	Paper Presentation 1 (20 min)
10:20-10:40	Paper Presentation 2 (20 min)
10:40-10:50	Coffee Break
10:50-11:10	Paper Presentation 3 (20 min)
11:10-11:20	Closing Remarks and Future Directions (10 min)
11:20-13:00	Poster Presentations
13:00	Lunch

Call for Papers

We invite submissions on topics including, but not limited to:

Ethical benchmarks for evaluating LLMs in Nordic and Baltic languages
Methods for creating culturally sensitive and inclusive evaluation datasets
Responsible techniques for generating or collecting alignment data
Challenges and solutions in ethical LLM alignment for less-resourced languages
Case studies on responsible LLM evaluation or alignment projects
Ethical considerations in LLM evaluation and alignment
Comparative studies of LLM performance and fairness in Nordic and Baltic languages
Innovative approaches to leveraging limited language resources in evaluation or alignment of language models

Submissions can be long papers (8 pages) or short papers (4 pages) in the ACL format. All submissions will undergo peer review by the program committee.

Submit Your Paper

Submission Portal

OpenReview.net/NoDaLiDa/Baltic-HLT/2025/Workshop/NB-REAL

Submit Your Paper

All submissions will be reviewed through a double-blind peer review process.

Please ensure your submission follows the NoDaLiDa format guidelines.

Important Dates

Workshop paper submission deadline: December 16, 2024
Notification of acceptance: January 13, 2025
Camera-ready papers due: February 3, 2025
Workshop date: March 2, 2025

Expected Outcomes and Impact

The NB-REAL workshop aims to:

Establish a collaborative network of researchers and practitioners.
Identify key challenges and opportunities in LLM evaluation and alignment for Nordic and Baltic languages.
Publish a collection of high-quality papers advancing the state of the art.
Formulate recommendations for ethical and sustainable practices in LLM evaluation and alignment.
Initiate an ongoing dialogue between researchers, industry professionals, and policymakers.

By addressing the unique challenges and opportunities in responsible LLM evaluation and alignment for Nordic and Baltic languages, NB-REAL will foster collaboration and advance the field in ways that benefit both the regional and global NLP communities.