<![CDATA[The rapid development of technology means that information is becoming more abundant and diverse, including in the movie industry. The number of movies each year can reach tens of thousands with various genres, making it difficult for potential viewers to choose a movie that suits their interests. One solution to the problem is the existence of a recommendation system that can provide movie recommendations based on information on movies that have been watched previously. Collaborative Filtering is a widely used approach in recommendation systems. Collaborative Filtering offers recommendations based on the similarity between users for User-based methods and the similarity between items for Item-based methods. However, the similarity value can be high for data with high dispersion even though only one item is in common. The Hybrid method can be a solution to overcome this by combining User-based and Item-based methods and adding genre information from the items. The final prediction result is obtained from the prediction results of all methods combined using linear combination. The combination is done by giving weight to each method and then summarising it. This study aims to determine the impact of weighting variations on Hybrid User-based and Item-based Collaborative Filtering methods. The results obtained from this study show that More Dominant User-based and Very Dominant User-based weightings are superior to other weightings because they show good performance for a smaller list of recommendations.]]>
Copyrights © 2024
