A developer has fine-tuned the Qwen2.5-7B language model on 100 films to generate probabilistic story graphs. This project, shared on Hacker News, aims to map narrative structures and character arcs from cinematic data. The model analyzes plot points to visualize how stories unfold, offering a new tool for understanding film composition.

This work sits at the intersection of AI and digital humanities. Large language models are increasingly used to analyze creative works, but applying them to structured narrative analysis is still emerging. By training on film scripts and summaries, the model learns patterns in storytelling that could aid writers, directors, and researchers studying narrative theory.

The approach raises questions about AI's role in creative analysis. While the model identifies patterns, it doesn't replace human interpretation of artistic intent. Future iterations could expand to more films or different media, potentially helping studios understand audience engagement or assist in script development. The community's feedback on Hacker News will likely shape its next steps.