| Popis: |
This work proposes a cluster-driven content-based recommender system for a video-sharing application. The system addresses the challenge of information overload and aims to enhance the user experience by providing personalized recommendations. The system leverages an online clustering technique to generate recommendations and incorporates a knowledge-based approach to recommend content to new platform users. The items in the system are embedded using Natural Language Processing techniques, and the evaluation metrics include diversity and unexpectedness. In addition, the system introduces a novel "exploration mode" that allows users to receive more diverse recommendations and discover new content. |
