Deriving Character Logic from Storyline as Codified Decision Trees

2026-01-15

AI Summary

Existing role-playing agents often exhibit behavioral inconsistency and fragility due to the absence of structured, executable, and empirically validated behavior specifications. This work proposes the Codified Decision Trees framework, which automatically induces context-sensitive, deterministic behaviors by extracting conditional-action rules from large-scale narrative data. The resulting hierarchical decision trees are interpretable, verifiable, and updatable, enabling robust and coherent agent behavior. By integrating iterative rule induction, data-driven validation, and hierarchical specialization, the method significantly outperforms both handcrafted and alternative automated approaches across 85 characters from 16 narrative works, markedly improving behavioral consistency, reliability, and transparency.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Agent-Based Simulation and Emergent Behavior

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Role-playing (RP) agents rely on behavioral profiles to act consistently across diverse narrative contexts, yet existing profiles are largely unstructured, non-executable, and weakly validated, leading to brittle agent behavior. We propose Codified Decision Trees (CDT), a data-driven framework that induces an executable and interpretable decision structure from large-scale narrative data. CDT represents behavioral profiles as a tree of conditional rules, where internal nodes correspond to validated scene conditions and leaves encode grounded behavioral statements, enabling deterministic retrieval of context-appropriate rules at execution time. The tree is learned by iteratively inducing candidate scene-action rules, validating them against data, and refining them through hierarchical specialization, yielding profiles that support transparent inspection and principled updates. Across multiple benchmarks, CDT substantially outperforms human-written profiles and prior profile induction methods on $85$ characters across $16$ artifacts, indicating that codified and validated behavioral representations lead to more reliable agent grounding.

Problem

• role-playing agents
• behavioral profiles
• narrative consistency
• executable representation
• agent grounding

Innovation

• Codified Decision Trees
• behavioral profiling
• narrative grounding
• interpretable AI
• role-playing agents