隨著OTT平台的快速發展，在平台上的資訊過於龐大，使用者需要花費額外的時間進行選擇和分析所需內容。對平台方而言，因為以訂閱制提供服務，所以維持使用者對平台的興趣，以避免使用者流失，因此使用推薦系統輔助消費者選擇。在推薦系統領域，混合過濾（Hybrid filtering）被廣泛應用，以解決單一推薦算法所帶來的問題。然而，當前大多數研究專注於推薦系統的準確性，對於提高多樣性的探討相對較少。因此，本研究結合兩種不同的推薦算法，專注於OTT平台電影推薦應用的研究。比較不同混合方式對推薦系統多樣性與準確性的影響。研究旨在探討利用協同過濾方法改善內容基礎過濾推薦多樣性的效果，並使用MovieLens與TMDB資料集所結合的資料作為實驗使用。最終，本研究將提出一個在多樣性和準確性方面平衡且具有應用價值的推薦系統，以提高OTT平台的使用體驗和用戶忠誠度。;With the rapid development of OTT platforms, users face the challenge of information overload, requiring additional time to select and analyze the desired content. Furthermore, many OTT platforms adopt subscription-based models, making the accuracy and diversity of recommendation systems crucial for maintaining user interest in the platform. In the field of recommendation systems, hybrid filtering is widely used to address the issues arising from employing a single recommendation algorithm. However, the majority of current research focuses on the accuracy of recommendation systems, with relatively less exploration on enhancing diversity. Therefore, this study combines two different recommendation algorithms, focusing on the application of movie recommendations on OTT platforms. Building upon collaborative filtering as the foundation, content-based filtering is used to extend the existing recommendation list, comparing the effects of different hybrid approaches on the accuracy and diversity of the recommendation system. The objective of this study is to explore the efficacy of enhancing the diversity in recommendations generated by content-based filtering by incorporating collaborative filtering techniques. Ultimately, this study propose a recommendation system that achieves a balance between accuracy and diversity, with broad applicability, to enhance the user experience and loyalty on OTT platforms.
