Narrative Theory-Driven LLM Methods for Automatic Story Generation and Understanding: A Survey

The survey discusses the integration of narrative theories with large language models (LLMs) for automatic story generation and understanding, highlighting the diverse sources of narrative texts and the current limitations in generation tasks compared to understanding. It emphasizes the need for theory-based metrics for narrative attributes, large-scale analysis, and situated narrative generation, advocating for a more systematic approach to narrative research in NLP. This work is significant for practitioners as it outlines potential directions for improving narrative quality and validation methods in LLM applications.