Short papers
Up to 4 pages of content, with unlimited additional pages for references, ethics/limitations, and supplementary material statements.
NLP/CORE Unranked
Industrial NLP 2026
Industrial NLP: NLP Research Towards Safety, Reliability, and Real-World Impact
The Workshop for Industrial NLP 2026 is a half-day event co-located with INLG 2026, focused on natural language generation (NLG) and large language model (LLM) research driven by real-world deployment challenges. It aims to bridge academia and industry by addressing safety, reliability, evaluation, and practical impact of deployed NLP systems across domains like healthcare, finance, and cybersecurity.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Extended abstracts
Up to 4 pages of content, with unlimited additional pages for references, ethics/limitations, and supplementary material statements.
Position papers
Up to 4 pages of content, with unlimited additional pages for references, ethics/limitations, and supplementary material statements.
Industrial case studies
Up to 4 pages of content, with unlimited additional pages for references, ethics/limitations, and supplementary material statements.
Research areas in scope
01
Safety & Security
Content safety, toxicity, and harmful-output mitigationAdversarial robustness, jailbreaks, and prompt injectionGuardrails and moderationPrivacy, PII leakage, and data protectionSecurity of NLP supply chains and toolingCompliance with safety regulations and standards
02
Reliability & Fairness
Factuality, hallucination detection, and groundingCalibration, uncertainty, and abstentionConsistency, prompt-sensitivity, and robustness to distribution shiftBias, fairness, and equitable behavior across users, dialects, and demographicsTrustworthy and accountable generation
03
Efficiency
Latency-, cost-, and memory-aware NLGModel compression, distillation, quantization, and pruningEfficient inference, caching, and serving at scaleSmall, on-device, and edge-deployable modelsSustainable and energy-aware NLP
04
Industrial Applications & Deployment
NLG/LLM systems in production across domains — including medicine and healthcare, finance, cybersecurity, law, education, search, assistants, customer support, and content platformsDeployment constraints (scalability, maintainability, integration)Domain adaptation and customizationMonitoring, incident analysis, and LLMOpsIntegration with enterprise and public-sector workflows
05
Evaluation, Datasets & Benchmarks
Evaluation in context: pipelines, protocols, and metrics that reflect real deploymentHuman evaluation and LLM-as-judge methodsPost-deployment quality assurance and regression testingNew datasets and benchmarks capturing operational conditions, domain shift, and multilingual/real-user settings
06
NLP for Social Good & Real-World Impact
NLG/LLM systems built for public-interest and high-benefit outcomesBroadening access through low-resource and multilingual NLPAccessibilityApplications in healthcare access, education equity, humanitarian and crisis response, public services, and the environmentParticipatory and human-centered design with affected communities
07
Negative Impacts & Their Mitigations
Critical analysis of the harms of deployed generative systems — misuse and dual-use, misinformation, labor and economic effects, environmental cost, and over-relianceConcrete, actionable mitigationsPosition papers, audits, incident retrospectives, and responsible-disclosure practices
Policies worth checking twice
- Submissions must be anonymised for double-blind review
- All submissions are non-archival and will not be published in the ACL Anthology proceedings
- Submissions may be previously published or under review elsewhere
- Submissions must use official ACL LaTeX/Word style files and follow ACL Author Guidelines
- Submissions must not exceed 4 pages of content (references, ethics/limitations, and supplementary material are unlimited)
- Submissions must follow the ACL Code of Ethics and include explicit ethical considerations and a data availability statement for experimental work
- Submissions are made through OpenReview or via the ARR Commitment track
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026