Leveraging Transformer Models for Predictive Analytics of Design Innovation Trajectories: A Cross- Disciplinary Approach to Market Success and Cultural Resonance

Abstract

Here we introduce a novel framework that adapts natural language processing techniques to analyze sequences of design innovation events, aiming to predict the market success and cultural impact of new products. By conceptualizing the key events in the design innovation process—from ideation and prototyping to market launch and user adoption—as a form of ‘design language’, we can capture the complex temporal dependencies and underlying patterns within these sequences. This study leverages large-scale, multi-source data, including design documentation, user engagement logs, market sales figures, and social media discourse, to construct comprehensive trajectories of design innovation. Our proposed model, Design2Vec, not only achieves high accuracy in predicting a product’s market performance and cultural resonance but also uncovers the critical design events and strategies that drive these outcomes. This work offers a cross-disciplinary perspective that integrates design studies, computer science, business management, and cultural studies, providing data-driven insights for future design practices and innovation strategies.