Book In A Bottle: Cross-Domain Book-Music Recommendations Using Hybrid Filtering and Sentiment Analysis

Traditional recommendation systems tend to focus more on following mainstream trends, neglecting niche preferences and cross-domain integration. Previous approaches prefer methods that exploit past user interactions or give priority to popular trends, overlooking cross-domain recommendations. Existing systems have neglected cross-domain recommendations as well. The goal of this system is to utilize sentiment analysis, collaborative and content-based filtering, and user-generated databases to create a hybrid recommendation system that provides personalized book recommendations paired with mood based playlists. Users are able to suggest books, which contributes to the development of a dynamic and varied database. This enables the system to cater to both popular as well as niche interests by providing unique recommendations. The proposed system is developed and tested on the Goodreads Books, Users and User Interaction datasets along with Music dataset and a custom dataset acquired via a structured survey. The results indicate the system’s effectiveness in recommending books and playlists in terms of accuracy, recall and precision.