The User, Plural

The User,
Plural

Heterogeneity and Interaction in User Simulation

A one-day workshop at COLING 2027 in Macau, China.

Read the call for papersImportant dates

Workshop date
TBD
Format
One day, hybrid
Location
Macau, China
COLING main conference
May 9–14, 2027

When, how, and for what purposes can simulated users serve as valid proxies for real users?

The User, Plural brings together researchers in natural language processing, human–computer interaction, behavioral science, psychology, and responsible AI around this central question.

User simulation offers a scalable way to generate training data, probe system behavior, discover failures, and support repeatable evaluation. Yet real users clarify, withhold information, correct misunderstandings, change goals, and abandon tasks. These behaviors, and the diversity of the people who exhibit them, are central to the workshop.

We focus on interactional fidelity—whether simulators preserve meaningful user behavior as an interaction unfolds—and user heterogeneity—whether they capture variation across language, culture, accessibility, expertise, preferences, and user state, including within-group and intersectional variation and rare behaviors.

Validity is fit-for-purpose. Evidence sufficient for generating data may not support ranking systems, drawing population-level conclusions, or informing consequential decisions. We ask how intended use should determine the human evidence, validation unit, uncertainty, and transfer assumptions needed to support a claim.

This first edition builds on the ACL 2026 User Simulation Birds-of-a-Feather session and its continuing working group. Together we aim to develop a shared research agenda and a practical reporting checklist covering simulator purpose, target users, validation, and limitations.

Four research directions

With particular attention to interactional fidelity and user heterogeneity.

  1. 1Capabilities and UsesWhat can simulated users reliably do?
  2. 2Human–Simulator GapsWhich differences matter?
  3. 3Evaluation and ValidationWhat evidence makes a simulator trustworthy?
  4. 4Responsible and Real-World UseWhen should simulation inform decisions?
The Macau waterfront seen from Taipa, with Macau Tower and Grand Lisboa.
Macau skyline (2013) by Mimihitam, CC BY-SA 3.0.

Announcements

  • The workshop has been accepted at COLING 2027.

  • The first call for workshop papers is scheduled for . Direct submissions are due ; pre-reviewed ARR commitments are due . The submission portal is TBD. See important dates and submission information.

  • We welcome reviewers with expertise in user simulation and related areas. See the reviewer call.

  • Questions: theuserplural@gmail.com