DEVS Formalism Enables Discrete-Event World Models from Natural Language

A novel method for constructing discrete-event world models from natural language descriptions has been introduced, utilizing the DEVS (Discrete Event System Specification) framework. This research seeks to bridge the divide between expensive, manually crafted simulators and neural models that face increasing inconsistencies over extended periods. By generating world models in real-time from natural language inputs, the approach aspires to merge the dependability of traditional simulators with the flexibility of neural networks. The study focuses on critical areas such as supply chains, procurement networks, and business processes, which are influenced by discrete events, timing constraints, and causal relationships. The paper can be found on arXiv with the ID 2603.03784.