Full Papers
Original submissions of 4–8 pages (excluding references) following the ACM CIKM 2026 template; no distinction between long and short papers.
AI/CORE Unranked
CIKM 2026
The 2nd Workshop on LLM Agents for Social Simulation @CIKM2026
The 2nd Workshop on LLM Agents for Social Simulation (LASS @ CIKM 2026) brings together researchers from AI, NLP, HCI, and vertical domains to advance the development of Large Language Model-based agents for simulating human behavior and social dynamics. The workshop focuses on fundamental challenges in modeling, evaluation, safety, and ethical deployment of LLM agents in real-world applications such as healthcare, education, and public governance, building on the success of LASS 2025.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Research areas in scope
01
Target Communities & Domains
AI & NLPHCISocial SciencePsychologyHealthcare & MedicineEducationPublic GovernanceEconomics & PolicyComputational Social Science
02
Knowledge & Mechanism Modeling
Incorporating domain knowledge, theories, and rulesCharacterizing persona, roles, preferences, and beliefsCapturing interaction and collective behaviorMemory dynamics, experience accumulation, self-evolution
03
Simulation Evaluation & Analysis
Evaluation criteria, rubrics, and protocolsExpert-in-the-loop evaluation, cost-effective annotationIntervention analysis of policies and treatmentsPhenomenon discovery and hypothesis generationHuman-agent interaction and behavioral impact analysis
04
Safety, Ethics & Accountability
Risk taxonomies, safety assurance and evaluationBias, fairness, stereotypes, and differential impactsPrivacy, consent, and sensitive data protectionEthical use of simulated human or community proxiesMisuse risks: manipulation, fraud, surveillance, etc.Transparency, auditability, and expert oversight
05
Benchmarks, Datasets & Infrastructure
Domain-specific datasets, benchmarks, evaluation suitesPlatforms, toolkits, APIs, and software systemsReproducibility standards and reporting protocols
Policies worth checking twice
- All submissions undergo double-blind peer review with at least three reviewers per paper and no conflicts of interest.
- Papers must be properly anonymized.
- Submissions must follow the ACM CIKM 2026 template and be 4–8 pages in length, excluding references.
- Submissions to other venues (e.g., ACL, EMNLP) are welcome and do not affect original venue eligibility.
- Accepted papers are eligible for fast-track recommendation to the Journal of Computer Science and Technology (JCST, CCF-B), subject to authors' consent.
- A Best LASS Paper Award will be presented at the closing ceremony with formal certification.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026