Full papers
Up to 8 pages for substantial contributions.
NLP/CORE Unranked
UncertaiNLP2026 2026
Third Workshop on Uncertainty-Aware NLP @ EMNLP 2026
UncertaiNLP2026 is the third workshop on Uncertainty-Aware NLP, co-located with EMNLP 2026 in Budapest, Hungary. It brings together researchers to explore how uncertainty in human language and NLP models—especially LLMs—can be formally represented, modeled, and leveraged to improve reliability, evaluation, and generation in natural language processing.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short papers
Up to 4 pages for ongoing or preliminary work.
Research areas in scope
01
Workshop Topics
Formal tools for uncertainty representationTheoretical work on probability and its generalizationsSymbolic representations of uncertaintyDocumenting sources of uncertaintyTheoretical underpinnings of linguistic sources of variationData collection (e.g., to document linguistic variability, multiple perspectives, etc.)Explicit representation of model uncertainty (e.g., parameter and/or hypothesis uncertainty, Bayesian NNs in NLU/NLG, verbalised uncertainty, feature density, external calibration modules)Disentangled representation of different sources of uncertainty (e.g., hierarchical models, prompting)Reducing uncertainty due to additional context (e.g. clarification questions, retrieval/API augmented models)Learning from single and/or multiple referencesGradient estimation in latent variable modelsProbabilistic inference (Theoretical and applied work on approximate inference e.g., variational inference, Langevin dynamics)Unbiased and asymptotically unbiased sampling algorithmsUtility-aware decoders and controllable generationSelective predictionActive learningStatistical evaluation of language modelsCalibration to interpretable notions of uncertainty (e.g., calibration error, conformal prediction)Evaluation of epistemic uncertaintyTheoretical and empirical study of hallucination phenomena in NLU/NLGDescribing, formalising, categorising hallucination phenomenaMethods for detecting and quantifying hallucinationsMitigation techniques including uncertainty-aware generation, retrieval-augmented methods, and controllable generationRelationship between specific kinds (or sources) of uncertainty and hallucination occurrence
Policies worth checking twice
- Submissions must be in PDF format and follow EMNLP 2026 formatting guidelines using official ACL style templates.
- Three submission types are accepted: direct archival, direct non-archival, and ARR commitment.
- Non-archival submissions are not published in proceedings but allow future submission to other venues.
- ARR commitment submissions must have prior ARR reviews and be linked via OpenReview; EMNLP industry track reviews are not accepted.
- Non-archival submissions can be concurrently submitted to EMNLP 2026 (including industry track) but not to other EMNLP co-located workshops.
- Camera-ready versions for archival papers may include one additional page to address reviewer comments.
- Every direct submission must appoint one or more eligible authors for reciprocal reviews (standard load: 3 papers).
- Reciprocal reviewers must hold a PhD or have co-authored at least one peer-reviewed paper at ACL/EMNLP/NAACL/EACL/COLING/NeurIPS/ICML/ICLR/AAAI/IJCAI or comparable venues.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026