.png){kind=icon}
.png){kind=icon}
Return
ProTriPlay: A trinity framework for professional interactive theater based on LLM
.png){kind=icon}
OA
AI
Y
H
M
Y
Y
DOI：10.1007/s40747-025-02173-4.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Interactive theater powered by large language models faces significant challenges, including limited player agency in plot development, restricted interaction variety, and a lack of personalized experiences. This study presents ProTriPlay, a trinity framework that addresses these limitations by mimicking professional theatrical production structures. The framework integrates three specialized agents: a director agent, which provides global narrative control and plot coordination; a screenwriter agent, which generates contextual scenes and environmental descriptions; and an actor agent, which manages character-specific interactions and dialogue. The system enables comprehensive player engagement through both character dialogue and environmental object interactions, while dynamically adapting plot developments based on real-time player decisions. A dual-pronged evaluation was conducted, combining subjective assessments from drama professionals and general users with an objective analysis against existing single-agent and dual-agent frameworks. Results show that ProTriPlay performs excellently. Subjective evaluations highlight the superior entertainment value and plot leadership, while objective tests confirm the framework’s advantages in scenery, narration, and scene transitions, validating the efficacy of the trinity agent architecture. The framework successfully maintains narrative coherence while allowing meaningful player influence on story progression. These findings establish a novel technical paradigm for large language model-based interactive theater systems, with practical applications spanning creative content development, actor rehearsal training, and immersive entertainment experiences.
Keywords:
Interactive theater
LLM
Multi-agent framework
Agent-based storytelling
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.6
Papers:
2.1K
Citations:
6.6K
Organization
Cited Papers