Streaming platforms are drowning in unstructured data—reviews, scripts, viewer comments—that current AI models cannot easily learn from. This project addresses a fundamental gap in how AI handles creative content. Movies and TV shows generate vast amounts of unstructured information—dialogue, plot summaries, audience reactions—that resists the tidy, numerical formats AI systems require for accurate predictions. Without structured data, AI models cannot reliably forecast which shows will engage viewers or why. If successful, the project will transform messy creative content into quantifiable metrics—measurable attributes of a film or series that AI can process. These metrics could then feed into predictive tools that help streaming services, producers, and distributors make better decisions about what to commission, promote, or recommend. The result would be a more efficient, data-driven complement to human expertise in the entertainment industry, reducing guesswork in a sector where billions of pounds hinge on audience engagement. The project is applied, not fundamental science. It builds directly on existing GPT-based AI advances, aiming to produce practical tools for industry stakeholders within the grant period.
View original technical description
This project emerges from the latest advancements in GPT-based AI to turn unstructured data from creative domains, specifically movies and TV shows. To train accurate AI models, it is essential to use structured data that can be easily learned by AI. However, movies and TV shows often present information in unstructured formats, making it difficult to integrate such data into AI systems effectively. The primary objective of this project is to transform unstructured content from movies and TV shows into structured, quantifiable metrics that provide deeper insights into the quality of actual streaming content. Furthermore, this project will explore the relationship between these structured metrics and consumer engagement. By understanding how various attributes of movies and TV shows influence viewer interaction, we can design tools that improve forecasting accuracy and decision-making processes for stakeholders in the entertainment industry. Ultimately, the structured metrics developed in this project will not only enhance predictive models but also contribute to the development of specialized AI systems for movie and TV analysis. These systems will complement and augment human expertise in creative domains, offering a more efficient and data-driven approach to evaluating and predicting the success of visual media. The host institution, Shanghai Theatre Academy, has many experts in the entertainment industry who will provide valuable expert opinions and form the foundation of these quantitative metrics. This expertise makes the Academy a perfect fit for this project.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know