A book recommendation system based on emotion analysis

Reading, a vital activity for personal development and intellectual enrichment, faces declining popularity in contemporary society. This research introduces an innovative book recommendation system that integrates emotion analysis to enhance reader engagement. Traditional systems, primarily reliant on historical preferences, often fail to accommodate fluctuating emotional states. To address this limitation, we developed a system that utilizes BERT (Bidirectional Encoder Representations from Transformers) for emotion analysis of user-generated diary entries that are written in Japanese, and leverages a large language model (LLM) to generate personalized book recommendations based on these emotional assessments. A user study was conducted, demonstrating the system's efficacy, with participants reporting positively for both emotion analysis and recommendation accuracy. These findings underscore the potential of emotion-aware book recommendation systems to personalize reading experiences and stimulate renewed interest in literature.