<![CDATA[This study aims to analyze user preferences for Action, Horror, and Romance film genres in video streaming services by implementing a Gemini AI-based Collaborative Filtering algorithm. Data were obtained from 1,017 respondents through an online survey using a 1–5 Likert scale. The research stages include data cleansing, calculating genre similarity using cosine similarity, and implementing an item-based Collaborative Filtering algorithm. Furthermore, Gemini AI embedding was applied, which is the process of transforming each genre into a high-dimensional numerical vector representation to more accurately capture semantic relationships between genres. The results show that Action is the most preferred genre, while the highest similarity score between genres was found between Horror and Romance. The developed recommendation system successfully mapped genre similarities and provided relevant viewing suggestions based on other users’ preferences. The system achieved an effectiveness rate of 62.38%. These findings can serve as a foundation for developing more adaptive and personalized recommendation systems in the future.]]>
Copyrights © 2025
