EvolvingWorld: An Open-Schema Framework for Co-Evolving Role-Play Agents and World Model in Interactive Literary World
Computation and Language
Summary
The authors created EvolvingWorld, a system to simulate stories where characters and their environment change together over time. Unlike older systems that only mimic fixed characters or single scenes, their model keeps track of character details and world changes across many scenes. The system has two main parts: one that manages characters and another that keeps the world’s state updated. They trained the system using a large dataset from 57 books and developed tests to measure how well it keeps stories consistent over long periods. Their experiments show that EvolvingWorld better maintains believable character and world evolution in interactive literary simulations.
Authors
Qing Zong, Yue Guo, Mengxin Yang, Yiwen Guo, Yangqiu Song
Abstract
This paper introduces EvolvingWorld, a framework and benchmark for character and world co-evolution in interactive literary worlds. Existing systems either treat interactive literary simulation as static persona imitation or isolated scene generation, failing to capture how characters and worlds evolve together over time. To address this, EvolvingWorld models literary simulation as a long-horizon process where characters interact, scenes progress, and character and world states are persistently updated. Unlike prior systems relying on fixed schemas, EvolvingWorld adopts an open-schema framework to support simulation across diverse literary worlds. The framework consists of two coupled modules: a Character Agent for multi-character role-play and persistent profile evolution, and an LLM-based World Model for global and location/entity-level state maintenance and scene progression. Based on this architecture, we formulate 7 trainable tasks for scene initialization, interaction generation, and state update. We construct a dataset from 57 books, producing 138,596 supervised training samples and 222 snapshots for testing. Furthermore, we introduce a trajectory-level LLM-as-Judge evaluation protocol spanning 10 dimensions and 20 metrics. Experiments show that EvolvingWorld can improve long-horizon simulation by effectively maintaining persistent, coherent character and world development.