Integrating Psychological and Subconscious Data into Recommender Systems: A Novel Model for Digital Advertising

Abstract

The growing complexity and volume of digital advertising have made recommender systems essential for enhancing user engagement and campaign performance. However, existing models predominantly rely on behavioral data, neglecting critical psychological and subconscious dimensions of user perception. This study introduces a novel hybrid recommender system that integrates multidimensional inputs, personality traits (Big Five model), subconscious associations (captured via ZMET), customer inspiration scores, and ad content tags, to deliver more psychologically aligned advertising recommendations. Using a sample of 549 participants exposed to four distinct ads from a pool of 625, data were collected through NEO personality inventories, inspiration scales, ZMET-based image selection, and expert ad tagging. The model was evaluated using standard classification metrics, achieving an accuracy of 91.5% and an AUC of 0.957, substantially outperforming conventional approaches. Key personality traits, especially Openness and Extraversion, were identified as significant predictors of recommendation relevance. This research demonstrates the value of combining behavioral, psychological, and subconscious data to build more intelligent, human-aware recommender systems. The findings offer practical insights for designing personalized ad campaigns and improving marketing efficacy in digital environments.