Sistem rekomendasi adalah sistem yang digunakan untuk merekomendasikan sesuatu item, misalnya merekomendasikan film. Banyaknya informasi film di internet membuat kita kesulitan dalam memilih film yang sesuai dengan selera. Untuk itu sistem rekomendasi film ini dibutuhkan oleh masyarakat. Sistem rekomendasi ini membutuhkan data rating dari movielens. Pemilihan metode sangat mempengaruhi keberhasilan sistem. Semakin akurat sistem yang dibuat maka semakin bagus. Metode yang populer saat ini adalah collaborative filtering khususnya metode collaborative filtering tradisional. Namun metode ini memiliki masalah skalabilitas sehingga untuk mengatasinya perlu ditambahkan metode clustering. Pada penelitian sebelumnya ada yang menggunakan metode k-means clustering dan user-based collaborative filtering. Collaborative filtering terdiri dari user-based dan item-based collaborative filtering. Pada penelitian lain menyebutkan bahwa metode item-based lebih akurat daripada metode user-based collaborative filtering. Untuk itu pada penelitian ini menggunakan metode k-means clustering dan item-based collaborative filtering. Hasil dari uji coba yang dilakukan menunjukkan bahwa metode pada penelitin ini lebih akurat daripada user-based dan item-based collaborative filtering. Presentase kenaikan akurasi metode yang diteliti dengan user-based adalah 113% sedangkan bila dibandingkan dengan item-based collaborative filtering kenaikan akurasinya adalah 168%. Metode yang diteliti juga lebih akurat dibandingkan metode gabungan user clustering dan user-based collaborative filtering dengan kenaikan akurasi sebesar 27%. Kata kunci: Film, Sistem Rekomendasi, Item Clustering, Item-Based Collaborative Filtering.

The recommendation system is a system used to recommend an item, for example recommending a film. The amount of information on films on the internet makes it difficult for us to choose films that suit our tastes. For this reason, the film recommendation system is needed by the community. This recommendation system requires rating data from movielens. The choice of method greatly influences the success of the system. The more accurate the system is made the better. The method that is popular now is collaborative filtering, especially traditional collaborative filtering methods. But this method has a scalability problem, so to overcome this, a clustering method is needed. In previous studies there were those using the k-means clustering and user-based collaborative filtering methods. Collaborative filtering consists of user-based and item-based collaborative filtering. In another study, the item-based method was more accurate than the user-based collaborative filtering method. For this reason the k-means clustering and item-based collaborative filtering methods are used in this study. The results of the trials conducted show that the method in this research is more accurate than user-based and item-based collaborative filtering. The percentage increase in accuracy of the user-based research method is 113% while when compared to item-based collaborative filtering the increase in accuracy is 168%. The method studied was also more accurate than the combined method of user clustering and user-based collaborative filtering with an increase in accuracy of 27%. Keywords: Film, Recommendation System, Item Clustering, Item-Based Collaborative Filtering.