NarrativePlay: Interactive Narrative Understanding

Runcong Zhao,Wenjia Zhang,Jiazheng Li,Lixing Zhu,Yanran Li,Yulan He,Lin Gui

2023-10-02

Abstract:In this paper, we introduce NarrativePlay, a novel system that allows users to role-play a fictional character and interact with other characters in narratives such as novels in an immersive environment. We leverage Large Language Models (LLMs) to generate human-like responses, guided by personality traits extracted from narratives. The system incorporates auto-generated visual display of narrative settings, character portraits, and character speech, greatly enhancing user experience. Our approach eschews predefined sandboxes, focusing instead on main storyline events extracted from narratives from the perspective of a user-selected character. NarrativePlay has been evaluated on two types of narratives, detective and adventure stories, where users can either explore the world or improve their favorability with the narrative characters through conversations.

Computation and Language,Artificial Intelligence,Human-Computer Interaction

What problem does this paper attempt to address?

The paper aims to address the issue of how to create an immersive interactive narrative system. Specifically, the researchers developed a new platform called NarrativePlay, which allows users to play the roles of characters in a novel and interact with other characters. To achieve this goal, the researchers utilized large language models (LLMs) to generate human-like responses based on character traits extracted from the novel. Additionally, the system integrates automatically generated visual displays of the narrative environment, including scene settings, character portraits, and dialogues, thereby greatly enhancing the user experience.

Compared to previous research, NarrativePlay focuses on extracting key events from the narrative rather than predefined sandbox environments. This approach simplifies the complexity of identifying the narrative environment and improves the system's adaptability. Furthermore, the researchers overcame the challenge of generating environments with reasonable detail by leveraging stable diffusion models as an external knowledge base. Finally, the researchers classified player behaviors and compiled common issues in interactive narratives to evaluate the quality of agent responses. Overall, the paper proposes a new method for creating immersive interactive narrative experiences, particularly in the genres of detective stories and adventure stories.