ANIME RECOMMENDATION CHATBOT USING HYBRID FILTERING & TRANSFORMERS WITH IMPACT OF GENRE DIVERSITY

This research paper proposes an anime recommendation chatbot that incorporates the aspects of filtering techniques as well as the transformer models to perform the recommendation session; however, the importance of the aspect of the diverse genre is stressed here. Anime itself is a peculiar kind of tv shows which has its own specific features; the mainstream shows satisfaction relation by genres is quite variable. This work combines the content-based and hybrid filtering models with the transformer model to know how diversification based on genre impacts the precision of the recommendations and user relevance. Making use of an extensive anime dataset containing not only genre and rating, but also user interaction indicators, we then compare the performance of each model. The chatbot goes further than recommending and providing alternatives and questions users to select genres and provide their feedback in order to improve them continuously. Consequently, findings show that our hybrid model in tandem with the use of variety in genres yields recommendation precision and user satisfaction levels above traditional practices. Also, the conversational component of the chatbot based on transformers increases the level of user’s satisfaction using the product due to the natural and individual approach to the dialogue. The present work stresses the significant of genre variety in media recommendations and recommends future research in implementing higher level of filtering and natural language processing to feedbacks for enhancing the recommender systems. Keywords— Content-based filtering, Hybrid Filtering, Genre Diversity, Recommendations, Transformers.