- Corpora - ELRA lists

Funded PhD openings in multi-agent AI safety, security, and privacy | University of Miami | Spring/Fall 2027
by Guy Zyskind 25 Sep '26

25 Sep '26

PI: Guy Zyskind I’m recruiting PhD students to join my lab at the University of Miami in Spring or Fall 2027. The web is moving toward AI agents interacting with other agents, making decisions and taking actions on our behalf. Our lab studies how these systems behave, how they fail, and what it takes to trust them with real responsibility. We are working in two broad directions: - Multi-agent AI safety, trust, and security. Studying agents that work together under one owner or interact on behalf of different people and organizations. We want to understand when these interactions produce harmful or unintended behavior, how to evaluate safety across the whole system, and how to detect and reduce risks while preserving agents’ capabilities. - Privacy-preserving AI, especially in healthcare and finance. Developing ways to run AI on sensitive data without exposing it. Examples include private inference and retrieval, and helping hospitals or financial institutions use information across organizations while keeping patient and customer records confidential. I’m looking for students who love research and want to make a lasting impact, whether through research work that shapes this new web or companies that change the world. You should have a record of pursuing ideas independently and want to help shape what we work on. You should have a strong foundation in at least one of: - AI safety or security - Data science, computational social science, or network science - Cryptography or computer security Full funding is available for strong students to pursue a PhD at the University of Miami, a leading U.S. R1 research university. Our lab collaborates on AI safety research with researchers at MIT, Stanford, and other leading institutions. Please email guyzyskind(a)miami.edu with your CV and a brief description of your research experience and interests.

1 0

Extended submission deadline: Swedish Language Technology Conference 2026 (SLTC 2026), Luleå Sweden
by Simon Dobnik 24 Sep '26

24 Sep '26

Please see below an email from György about an extended deadline to submit abstract to SLTC 2026 in Lulelå: https://www.ltu.se/en/research/research-subjects/machine-learning/the-swedi… Best regards, Simon — Simon Dobnik Professor of Computational Linguistics CLASP & FLoV, University of Gothenburg https://www.gu.se/en/about/find-staff/simondobnik Begin forwarded message: From: György Kovács <gyorgy.kovacs(a)ltu.se> Subject: Re: Cfp: The Second Workshop on Ethics for Research and Teaching in Natural Language Processing (Ethics4NLP-2026) +events Date: 21 September 2026 at 18:29:26 GMT+2 To: Simon Dobnik <simon.dobnik(a)gu.se> Over the past few days, we have received many emails from PhD students and researchers from across Sweden asking whether we could extend the submission deadline. We heard you! 😊 The new submission deadline for the Swedish Language Technology Conference 2026 (SLTC 2026) is: 📅 September 26, 2026, 23:59 AoE So, if you were hoping to submit but needed a little more time, you now have another opportunity to share your work with the Swedish language technology community! Submit your work, join us in Luleå on November 23–24, and combine research and networking with a visit to the beautiful north of Sweden. And November in Luleå has something extra to offer: snow, the possibility of seeing the magical Northern Lights, and, depending on winter conditions, perhaps even the chance to experience Luleå's famous ice road. So… you have a few extra days. We hope to see your submission — and to welcome you to Luleå in November! 😊 📌 New submission deadline: September 26, 2026, 23:59 AoE More information about the conference and submissions is available here<https://www.ltu.se/en/research/research-subjects/machine-learning/the-swedi…>: https://www.ltu.se/en/research/research-subjects/machine-learning/the-swedi…

1 0

Call for workshop proposals: NoDaLiDa 2027
by Jenny Kunz 24 Sep '26

24 Sep '26

We are glad to announce that the 26th Nordic Conference on Computational Linguistics (NoDaLiDa 2027) will be held in Copenhagen, Denmark, on May 25–28, 2027. We invite proposals for workshops to be held in conjunction with NoDaLiDa 2027. Workshops can be scheduled either for a full day (morning and afternoon) or for half a day. The workshops will be held on Tuesday, May 25, and Friday, May 28. NoDaLiDa addresses all aspects of natural language processing, speech processing, and computational linguistics, including work in closely related neighboring disciplines (such as, for example, machine learning, linguistics, digital humanities, or psychology) that is sufficiently formalized or applied to bear relevance to speech and language technologies. The main conference will be held on-site only, without an online option, in order to facilitate networking. Workshops are free to offer online presentations if they wish to do so. Workshop proposals can be submitted through this form: https://forms.gle/pLFbGKKb8siG6E1T8 Workshop proposals must include adequate information on the following aspects: * A title, short name / acronym, and a brief description of the workshop topic and content. * Workshop organizer(s) and contact information * If the workshop has been held before, details on: how many prior editions occurred, where previous workshops were held, how many papers were accepted, and an estimate of how many in-person attendees the workshop attracted * An estimate of the maximum number of attendees of the planned edition * Duration of the workshop (full or half-day) * Mode of organization (on-site only or hybrid) * Date preferences IMPORTANT DATES * November 16, 2026: Submission of workshop proposals * December 7, 2026: Notification of workshop selection TENTATIVE WORKSHOP TIMELINES * First call for workshop papers: December 14, 2026 * Second call for workshop papers: January 18, 2027 * Workshop paper submission deadline: March 15, 2027 * Notification of acceptance: April 12, 2027 * Camera-ready paper due: April 26, 2027 * Workshop proceedings due: May 3, 2027 * Workshop dates: May 25 and 28, 2027 WORKSHOP CHAIRS * Maria Barrett, AMD Silo AI, Denmark * Shaoxiong Ji, University of Turku, Finland

1 0

Open Positions (2x PhD, 1x Postdoc) at LMU Munich on CSS, Political Psychology, Interventions on Democracy
by Wuttke, Alexander 23 Sep '26

23 Sep '26

Hello everyone, We invite applications for three academic positions across two DFG-funded research projects: "Democratic Persuasion: How to Make the Case for Democracy" Using survey experiments, LLM-based persuasive conversations, and a framed field experiment involving one-on-one dialogues, this project identifies the motivations behind democratic skepticism and tests which communicative strategies reinforce support for democracy Emmy Noether Research Group: “Predictably Paradoxical - Mapping the Democratic Mind with AI" Developing and employing the novel method of AI Conversational Interviewing, this project analyzes multi-modal data from LLM-enabled semi-structured interviews to map people’s democratic belief systems We are hiring: 1. PhD Researcher working in both projects (75%, 4 years, TVL E-13, approx. €44,500/year) Link to Job Ad: https://www.dropbox.com/scl/fi/gkypvd8pgaxvxgtxrartd/Job_Ad_PhD_2027-CSS.do… 2. PhD Researcher in Computational Social Science (75%, 4 years, TVL E-13, approx. €44,500/year) Link to Job Ad: https://www.dropbox.com/scl/fi/9gscd16e88k395ss9w7gq/Job_Ad_PhD_Combi_2027.… 3. Postdoctoral Researcher on Democratic Persuasion (100%, fixed term: 22 months, TVL E-13, approx. €64,000/year) Link to Job Ad: https://www.dropbox.com/scl/fi/4vcicqtwv0s403sfk9cpc/Job_Ad_PostDoc_2027.do… Positions start in January 2027 or as soon as possible We offer: * A welcoming, inclusive environment at a leading European research university * Involvement in impactful, cutting-edge research * Dedicated resources for professional growth and research activities * Excellent quality of life in Munich Ideal candidates have a background in quantitative methods, proficiency in statistical programming (R and/or Python), and an interest in political psychology, persuasion, or democracy research. Applications will be reviewed on a rolling basis (until October 20th). So, the earlier you apply, the better. Prof. Dr. Alexander Wuttke Assistant Professor Digitalization and Political Behavior LMU Munich https://www.political-behavior.digital/

1 0

CLARIN ERIC Newsflash September 2026
by Gorgaini, E. (Elisa) 23 Sep '26

23 Sep '26

Dear colleagues, The September edition of the CLARIN Newsflash is out! Highlights include: * The publication of CLARIN2026 proceedings * The publication of Tour de CLARIN Volume VI dedicated to young scholars * Kick-off of new European projects * A visit of CLARIN-PL to the chancellery of the Prime Minister of Poland * Calls and opportunities to collaborate and connect Read the newsletter: https://www.clarin.eu/content/clarin-newsflash-september-2026 Subscribe to the newsletter: here<https://clarin.us12.list-manage.com/subscribe?u=8bae79dc9813797e81387ebaa&i…>. Kind regards, --- Elisa Gorgaini Communication Officer - CLARIN ERIC Utrecht University | Drift 10, 3512 BS Utrecht, The Netherlands e.gorgaini(a)uu.nl<mailto:e.gorgaini@uu.nl> | elisa(a)clarin.eu<mailto:elisa@clarin.eu> www.clarin.eu<https://www.clarin.eu>

1 0

NAIMES: AI-generated name translation from the ParlaMint 5.0 corpus
by Fernando Sánchez Rodas 23 Sep '26

23 Sep '26

NAIMES (Naming Artificial Intelligence Metric Evaluation System) is a postdoctoral Marie Curie research project examining the reliability of AI-generated translations, with a particular focus on the translation of proper names. The project uses parliamentary discourse from ParlaMint 5.0, a multilingual corpus of European parliamentary proceedings, as its source material. Such institutional texts contain names of people, or anthroponyms (e.g. Keir Starmer); names of places, or toponyms (e.g. Spain, Berlin); and names of organisations, events, and products, known as chrematonyms (e.g. Ministry of Foreign Affairs, the Istanbul Convention). Proper names are closely tied to language, culture, and society, and their forms and meanings may evolve and vary across contexts. As a result, the translation of proper names raises specific challenges for AI-based translation systems. The aim of NAIMES is to evaluate and compare how ChatGPT translates proper names from English, German, and Czech into Spanish, and to identify recurring translation patterns. The project has now entered its publication and dissemination phase. Upcoming activities include workshops on onomastics and language technologies at Palacký University Olomouc (together with project supervisor Dr. Enrique Gutiérrez Rubio, Department of Romance Languages) and Leipzig University (with Prof. Dr. Carsten Sinner). NAIMES is carried out at Palacký University Olomouc, Czech Republic, and is co-funded by the European Union and the Czech Ministry of Education, Youth and Sports through the OP JAC Project “MSCA Fellowships at Palacký University IV” CZ.02.01.01/00/22_010/0013054. For more information, please visit our dedicated links: - NAIMES website (https://fernandosanchezrodas.owlstown.net/pages/naimes-project-introduction) - Instagram (https://www.instagram.com/fsanchezrodas.phd) - X/Twitter (https://twitter.com/fsanchezrodas_)

1 0

CfP: 26th Nordic Conference on Computational Linguistics (NoDaLiDa 2027), Copenhagen, May 25–28, 2027
by Jenny Kunz 23 Sep '26

23 Sep '26

We are glad to announce that the 26th Nordic Conference on Computational Linguistics (NoDaLiDa 2027) will be held in Copenhagen, Denmark, on May 25–28, 2027. nodalida2027.ku.dk<http://nodalida2027.ku.dk/> SUBMISSIONS NoDaLiDa 2027 addresses all aspects of natural language processing, speech recognition and synthesis, and computational linguistics, including work in closely related neighboring disciplines (such as, for example, machine learning, linguistics, digital humanities, or psychology) that is sufficiently formalized or applied to bear relevance to speech and language technologies. We invite paper submissions of three types: * regular papers on substantial, original, and unpublished research, including empirical evaluation results, where appropriate; * short papers on smaller, focused contributions, work in progress, negative results, surveys, or opinion pieces; and * demonstration papers on software or resource demonstrations, e.g. of systems, interfaces, infrastructures, data collections, or annotations. We particularly encourage submission of papers on completed or ongoing work, where the first author is a Master’s or PhD student. This should be indicated at submission time. Papers accepted for presentation at the conference will be included in the NoDaLiDa 2027 proceedings, which are published in the ACL Anthology and the NEALT Proceedings Series at DSpace at Tartu University Library. IMPORTANT DATES * Monday, January 18, 2027: Direct Submission of Papers * Monday, February 22, 2027: Commitment of ARR-Reviewed Papers * Monday, March 8, 2027: Notification of Acceptance * Monday, April 5, 2027: Camera-Ready Manuscripts * Wednesday–Thursday, May 26–27, 2027: NoDaLiDa Main Conference The main conference will be held on-site only, without an online option, in order to facilitate networking. SUBMISSION FORMATS All submissions must follow the ACL style files (available at https://github.com/acl-org/acl-style-files ) and be submitted through the OpenReview conference management system. Submission links will be published with the second call for papers in October 2026. The page limits for submissions are: up to eight pages for regular papers and up to four pages for short papers and demo papers. For all three submission types, these page limits do not include additional pages with bibliographic references, as well as optional sections for limitations and ethical considerations. Appendices are allowed, but reviewers will not be required to consult them. Submissions must be anonymous, i.e., not reveal author(s) on the title page or through self-references. DOUBLE SUBMISSION AND PRE-PUBLICATION Papers that have been or will be submitted to other venues must indicate this at submission time and must be withdrawn from the other venues if accepted to NoDALiDa 2027. Authors of papers accepted for presentation at NoDALiDa must notify the program chairs by the camera-ready deadline as to whether the paper will be presented. We will not accept for publication or presentation papers that overlap significantly in content or results with papers that will be (or have been) published elsewhere. POLICY ON AI ASSISTANCE We follow the ACL policy for AI assistance: https://2023.aclweb.org/blog/ACL-2023-policy/. Submissions that include hallucinated references or appear to be AI generated to a degree that harms trust or readability will be desk rejected. PROGRAM COMMITTEE General Chair * Yves Scherrer, University of Oslo, Norway Program Chairs * Jindřich Helcl, University of Oslo, Norway * Hyewon Jang, University of Gothenburg, Sweden * Jenny Kunz, Linköping University, Sweden * Kairit Sirts, University of Tartu, Estonia Workshop Chairs * Maria Barrett, AMD Silo, Denmark * Shaoxiong Ji, University of Turku, Finland Publication Chair * Timothee Mickus, University of Helsinki, Finland Publicity and Social Media Chair * Jenny Kunz, Linköping University, Sweden Local Chairs * Ali Basirat, University of Copenhagen, Denmark * Costanza Navarretta, University of Copenhagen, Denmark * Sussi Olsen, University of Copenhagen, Denmark To inquire about the submission and reviewing process or the scientific program of the conference, or any other practical inquiries, please email nodalida-2027(a)googlegroups.com<mailto:nodalida-2027@googlegroups.com>. Follow us: * On LinkedIn<https://www.linkedin.com/company/nodalida>: https://www.linkedin.com/company/nodalida * on BlueSky<https://bsky.app/profile/nodalida.bsky.social>: @nodalida.bsky.social <https://bsky.app/profile/nodalida.bsky.social> * or on X: NoDaLiDa (@NoDaLiDa) / X<https://x.com/NoDaLiDa> Web page: nodalida2027.ku.dk<http://nodalida2027.ku.dk/>

1 0

[DEADLINE EXTENSION] Last call for papers for NLP4CALL 2026
by Arianna Masciolini 22 Sep '26

22 Sep '26

|[DEADLINE EXTENSION] Last call for papers for NLP4CALL 2026 November 19-20, 2026. Gothenburg, Sweden and online https://nlp4call.github.io/current/ ________________________________________________________________________________ We are happy to announce the 15th edition of the NLP4CALL workshop on Natural Language Processing for Computer-Assisted Language Learning. This year, NLP4CALL will be a two-day hybrid event that will take place on November 19-20, 2026. Onsite participation will be possible at the Humanisten building of the University of Gothenburg (Renströmsgatan 6, Göteborg, Sweden) and a link for remote participation will be shared closer to the workshop dates. Both onsite and online participation are free of charge. Contents: - Workshop description - Submission information - Important dates - Registration - Invited speakers - Organizers - Contact - Funding Workshop description ________________________________________________________________________________ The workshop series on Natural Language Processing (NLP) for Computer-Assisted Language Learning (NLP4CALL) is a meeting place for researchers working on integrating Natural Language Processing and Speech Technologies in CALL systems and exploring the theoretical and methodological issues arising from this connection. The latter includes, among others, the integration of insights from Second Language Acquisition (SLA) research, and the promotion of “Computational SLA” through setting up Second Language research infrastructures. The intersection of Natural Language Processing (or Language Technology / Computational Linguistics) and Speech Technology with Computer-Assisted Language Learning (CALL) brings “understanding” of language to CALL tools, thus making CALL intelligent. This fact has given the name for this area of research — Intelligent CALL, or short, ICALL. As the definition suggests, apart from having excellent knowledge of Natural Language Processing and/or Speech Technology, ICALL researchers need good insights into SLA theories and practices, as well as knowledge of second language pedagogy and didactics. This workshop therefore invites a wide range of ICALL-relevant research, including studies where NLP-enriched tools are used for testing SLA and pedagogical theories, and vice versa, where SLA theories, pedagogical practices or empirical data are modeled in ICALL tools. The NLP4CALL workshop series is aimed at bringing together competences from these areas for sharing experiences and brainstorming around the future of the field. We welcome papers: - that describe research directly aimed at ICALL; - that describe the ongoing development of resources and tools with potential use in ICALL, either directly in interactive applications, or indirectly in materials, application, or curriculum development, e.g. learning material generation, assessment of learner texts and responses, individualized learning solutions, provision of feedback; - that discuss challenges and/or research agenda for ICALL; - that describe empirical studies on language learner data. This year, the workshop has a special focus on process-oriented approaches to educational NLP, including but not limited to work related to the collection and analysis of keystroke data from langauge learners. Submission information ________________________________________________________________________________ Submissions should describe original unpublished complete or in-progress work and follow the ACL Guidelines for Generative Assistance in Authorship (https://www.aclweb.org/adminwiki/index.php/ACL_Policy_on_Publication_Ethics…). Each submitted paper will be peer-reviewed by at least two members of the program committee in a double-blind fashion. All accepted papers will be collected into a proceedings volume to be published both in the NEALT Proceeding Series and through the ACL anthology. We accept short, long and demo papers, all of which have to adhere to the following page limits: - short and demo papers: between 4 and 7 pages - long papers: between 8 and 12 pages. Also note that: - references and appendices do not count towards these page limits - the main body of the paper has to be self-contained, as reviewers are not required to read any appendices - camera-ready versions of accepted papers will be allowed an additional page to address reviewer comments. Papers should be submitted as PDFs through EasyChair (https://easychair.org/account/signin?l=7836699869611025050.1789051827.85f55…). LaTeX and Word templates are available at https://github.com/NLP4CALL/nlp4call.github.io/blob/website/_includes/other…. Important dates ________________________________________________________________________________ - paper submission deadline: September 27, 2026 (EXTENDED) - notification of acceptance: October 15, 2026 - camera-ready deadline: October 29, 2026 - workshop: November 19-20, 2026 All deadlines are AoE. Registration ________________________________________________________________________________ Registrations are open! You can find the registration form at https://forms.cloud.microsoft/pages/responsepage.aspx?id=Me2YB7D1NUmGPHPuJQ…. Invited speakers ________________________________________________________________________________ Prof. Andrea Horbach: A Focus on the Writing Process With and Without AI Support: Insights from the German PISA 2025 Foreign Language Assessment Data Collection Bio: Andrea Horbach has been professor for Educational NLP at Kiel University and the Leibniz Institute for Science and Mathematics Education (IPN) since 2024. Her research interests include free-text scoring, argument mining, automated feedback, and the use of process data for assessment. She is particularly interested in developing explainable and fair methods that support human agency in writing and assessment processes. Abstract: Automated essay scoring has traditionally focused on the text as the final written product. More recently, corpora that include keylogging data have made it possible to shift the focus toward writing as a process: to investigate how students write and revise their texts, and to explore whether scoring models can provide useful feedback before the final essay is completed. Moreover, AI-based writing support is becoming increasingly available, but we still know comparatively little about how students actually use such systems while writing. In this talk, I will give an overview of ongoing work based on data from the German PISA 2025 foreign language assessment. The data combine final essays and their scores with intermediate text states, keylogging information, and interaction logs from an LLM-based writing support chatbot. I will present first results on how intermediate texts and process information can be used as early evidence for automated scoring, and how students interact with AI support during writing. I will conclude by discussing what these data can tell us about timely feedback, the interpretation of process evidence, and the role of AI support in computer-assisted language learning. Dr. Rianne Conijn: Moving beyond the product: Keystroke Logging for CALL Bio: Rianne Conijn is an assistant professor in learning analytics in the Human-Technology Interaction group at Eindhoven University of Technology, the Netherlands. Her research focuses on developing methods to capture and model learning behavior (e.g., self-regulated learning) and writing behavior (e.g., revision processes), using fine-grained behavioral data, such as keystroke and clickstream data. In addition, she uses these methods to examine how educational technologies, such as generative AI, affect the learning process. Rianne recently received a NWO Veni grant on “Human-Centered AI in education” where she aims to improve human-AI collaboration in writing in higher education. Abstract: This abstract has been rewritten by me and polished with AI several times, yet you only read this final version. The final product of writing reveals little of the cognitive and behavioural activities involved in producing it. Keystroke logging, e.g. using Inputlog, can provide detailed insights into how writing unfolds. However, its value depends on how we interpret and analyse these data. In this talk, I show how keystroke data can inform our understanding of students’ writing at three levels, each making different aspects of the process visible and each requiring its own analytical choices. The micro level covers individual key events and short sequences, for instance showing that pauses cannot simply be equated with planning. The meso level characterizes how a writing session unfolds over time, revealing patterns that can inform writing instruction that whole-session averages obscure. The macro level extends the analysis across a project, tracing activity across sessions, drafts, and changing task demands. Finally, I discuss how these perspectives can help us investigate writing with generative AI, drawing on ongoing research into students’ master thesis writing. Together, these perspectives show how analysing writing at different levels can deepen our understanding of how students produce texts, both with and without generative AI. Organizers ________________________________________________________________________________ - Arianna Masciolini, Språkbanken Text, University of Gothenburg, Sweden - David Alfter, Gothenburg Research Infrastructure in Digital Humanities (GRIDH), University of Gothenburg, Sweden - Herbert Lange, Språkbanken Text, University of Gothenburg, Sweden - Ricardo Muñoz Sánchez, Språkbanken Text, University of Gothenburg, Sweden - Ildikó Pilán, Norwegian Computing Center, Norway - Elena Volodina, Språkbanken Text, University of Gothenburg, Sweden Contact ________________________________________________________________________________ For more information about the workshop, you are welcome to reach out to nlp4call2026(a)easychair.org. Funding ________________________________________________________________________________ This workshop is jointly supported by: - Språkbanken – jointly funded by its 10 partner institutions and the Swedish Research Council (2025–2028; project id 2023-00161); - Huminfra; - the Department of Swedish, Multilingualism, Language Technology; - the IncluEdS project. |

1 0

annotrAIn: A Platform That Simplifies Annotation Projects
by George Giannakopoulos 22 Sep '26

22 Sep '26

Dear all, I am sharing an interesting annotation project from SciFY <https://scify.org/en/> (a multiply awarded <https://scify.org/en/awards-distinctions/>, not-for-profit IT company I co-founded back in 2012). The project is called annotrAIn. More info and contact information below. Best regards, George G. > Anyone who has worked on annotation projects (e.g. to train AI models) knows > how much effort and coordination they require... > > *annotrAIn* is a new online platform developed by the award-winning non-profit > organisation Science For You – SciFY <https://scify.org/>. It brings the > entire annotation project workflow into a single environment: from project > design and task assignment to team communication, quality monitoring, and > final data export. > > Built on SciFY’s years of hands-on experience with real-world annotation > projects, the platform is designed for: > > * research teams developing AI models > * data science professionals and ML engineers > * teams managing annotation projects > * annotators and domain experts > > *What your team gains* > > * significantly lower administrative overhead, with information no longer > scattered across folders, emails, and different tools > * better coordination and easier task assignment > * more consistent annotated data, with errors identified early > * faster project completion > > The platform is bilingual (Greek/English) and has been designed with a strong > focus on accessibility. > > 👉 *Watch a short presentation of the platform > <https://go.scify.org/AnnotrAInPresentation>* > > *Would you like to try it?* > Contact us at *info(a)scify.org* to request a demo or run a pilot annotation > project using your own data. > -- -------------------------------------------------------------------------------- My working day may not be your working day. Please do not feel obliged to reply to this email outside of your normal working hours. -------------------------------------------------------------------------------- *George Giannakopoulos, PhD* /Researcher/ Home page <http://www.iit.demokritos.gr/~ggianna> SKEL Lab - NCSR Demokritos <http://www.iit.demokritos.gr> and /Scientific Officer/ ahedd DIH - NCSR "Demokritos" <https://ahedd.demokritos.gr> and /Co-founder, Chief Executive Officer/ SciFY Not-for-Profit Company <http://www.scify.org>

1 0

Second Call for Papers on Challenges in Pretraining
by Marco Antonio Stranisci 22 Sep '26

22 Sep '26

-------------------------------------------------------------------------------------- Second Call for Papers for the Special Track on Challenges in Pretraining -------------------------------------------------------------------------------------- Deadline: 31st October 2026 Website: https://jair.org/index.php/jair/specialtrack_pretrain Contact: marcos(a)itu.dk OVERVIEW -------------------------------------------------------------------------------------- The invention of Large Language Models (LLM) determined a paradigm shift in machine learning, which changed its focus from task-specific training of systems to their self-supervised pretraining on large data resources. This turn had a broad impact in Artificial Intelligence (AI) research, as it raises new challenges in the design and development of Machine Learning systems. Whilst the growing interest in language modeling is driving rapid innovation, research on pretraining remains fragmented. Training LLMs is a highly intersectoral endeavor that raises research challenges spanning a wide range of topics, addressed by multiple communities involved in the design of these technologies. However, the resulting research outputs are scattered across many venues, limiting a comprehensive understanding of the general challenges related to pretraining. The special track addresses this gap seeking submissions on the state of the art and existing challenges on pretraining. Specifically, the track focuses on the following broad content areas: ● *Pretraining Datasets*. The quality of pretraining datasets is crucial for developing LLMs. Recent research focuses not just on larger datasets but on data selection strategies aimed to improve performance efficiently. Data quality is also tied to governance in pretraining pipelines, requiring strong stewardship and archival practices. Methods like active learning and data minimization help reduce the amount of data needed while maintaining high-quality pretraining. ● *Model Architectures*. A large body of research explores what and how LLMs learn during pretraining, including studies on uncertainty, multilingual and token-free learning, and Mixture of Experts methods. It also examines architectural innovations: although autoregressive models dominate, alternatives such as neuro-symbolic and spiking neural networks aim to create more cognitively inspired learning approaches. ● *Computational Resources*. The growing computational need of training LLMs strongly influences pretraining strategies. Approaches like federated learning support distributed training while protecting data ownership and shaping model representations. Efficiently distributing computational loads to possibly heterogeneous platforms remains a complex optimization problem itself while alternative methods focus on training models under constrained computational budgets. This special track seeks contributions on the following, non-exhaustive list of topics: ● Advanced data selection and filtering strategies for pretraining ● Data stewardship and archival practices of pretraining data ● Data minimization and Active Learning approaches ● Neurosymbolic and cognitively inspired learning architectures ● Quantification and analysis of uncertainty in pretraining ● Tokenizer-free and alternative input representation approaches ● Distributed and federated learning for large-scale pretraining ● Language modeling under low-computational settings ● Benchmarking and resource efficiency analyses in pretraining The special track welcomes submissions that introduce innovative methodological contributions relevant to its topics. Papers that primarily rely on standard methods or established benchmarks fall outside the scope of this track. KEY DATES -------------------------------------------------------------------------------------- JAIR special tracks have a submission window. Papers can be submitted anytime during that window. They are reviewed as they arrive and accepted papers will go to production and will be published asynchronously as soon as they are ready, first as part of a usual JAIR pipeline and later on a JAIR webpage dedicated to the special track. *Target timeline*: ● Submission period: April 30, 2026 - October 31, 2026 ● First round of review and authors' notification: December 2026 ● Resubmissions: February 2027 ● Second round of review and authors' notification: April 2027 ● Final manuscripts: June 2027 Accepted submissions will be added to this page on publication. TRACK EDITORS -------------------------------------------------------------------------------------- Golnoosh Farnadi, McGill University Ferdinando Fioretto, University of Virginia Lucie Flek, University of Bonn Raphael Fischer, TU Dortmund Marco Antonio Stranisci, IT University of Copenhagen ---- Marco personal website: https://marcostranisci.github.io/

1 0

2026

2025

2024

2023

2022