Muhammed SÜTÇÜ, Ecem KAYA, Oğuzkan ERDEM

Movie Recommendation Systems Based on Collaborative Filtering: A Case Study on Netflix

Filmler, şarkılar ve alışveriş ürünleri gibi ögelerin kullanıcı değerlendirmeleri Öneri Sistemleri (ÖS) tarafından henüz değerlendirilmemiş ürünleri tahmin etmek için kullanılır. ÖS kullanıcılara çeşitli alanlarda öneri vermek için geliştirilmiştir ve ÖS uygulama alanlarından birisi de film önerisidir. Bu alanda üç genel algoritma kullanılmaktadır; kullanıcılar arası benzerliğe dayanarak tavsiye veren İşbirlikçi Filtreleme, kullanıcı-eşya eşleştirilmesindeki ilişkiden beslenen İçerik Tabanlı Filtreleme ve bu iki algoritmayı birleştiren Hibrit Filtreleme. Bu çalışmamızda İşbirlikçi Filtreleme çerçevesinde hangi metotların daha etkili çalıştığı incelenmiştir. Analizimizde Netflix Ödül veri seti kullanılmış ve iyi bilinen İşbirlikçi Filtreleme metotları olan Tekil Değer Ayrışımı, Tekil Değer Ayrışımı++, K En Yakın Komşu ve Eş Kümeleme kıyaslanmıştır. Her metodun hatası Ortalama Hata Kare Kökü kullanılarak ölçülmüştür. Son olarak, K En Yakın Komşu metodunun veri setimizde daha başarılı olduğu sonuçlanmıştır.

Anahtar Kelimeler:

film öneri, öneri sistemleri, işbirlikçi filtreleme, Netflix ödül

Movie Recommendation Systems Based on Collaborative Filtering: A Case Study on Netflix

User ratings on items like movies, songs and shopping products are used by Recommendation Systems (RS) to predict user preferences for items that have not been rated. RS has been utilized to give suggestions to users in various domains and one of the applications of RS is movie recommendation. In this domain, three general algorithms are applied; Collaborative Filtering that provides prediction based on similarities among users, Content-Based Filtering that is fed from the relation between item-user pairs and Hybrid Filtering one which combines these two algorithms. In this paper we discuss which methods are more efficient in movie recommendation under the framework of Collaborative Filtering. In our analysis we use Netflix Prize dataset, and we compare well-known Collaborative Filtering methods which are Singular Value Decomposition, Singular Value Decomposition++, K-Nearest Neighbour and Co-Clustering. The error of each method is measured by using Root Mean Square Error (RMSE). Finally, we conclude that K-Nearest Neighbour method is more successful in our dataset.

Keywords:

movie recommendation, recommendation systems, collaborative filtering, Netflix prize,

PDF

___

[1] W. Deng, R. Patil, L. Najjar, Y. Shi, and Z. Chen, “Incorporating Community Detection and Clustering Techniques into Collaborative Filtering Model,” Procedia Comput. Sci., vol. 31, pp. 66–74, 2014, doi: 10.1016/j.procs.2014.05.246.
[2] J. Bobadilla, F. Serradilla, and A. Hernando, “Collaborative filtering adapted to recommender systems of E-learning,” Knowledge-Based Syst., vol. 22, pp. 261–265, May 2009, doi: 10.1016/j.knosys.2009.01.008.
[3] I. Tobías, “Matrix factorization models for cross-domain recommendation: Addressing the cold start in collaborative filtering,” 2017.
[4] T. Anwar and V. Uma, “Comparative study of recommender system approaches and movie recommendation using collaborative filtering,” Int. J. Syst. Assur. Eng. Manag., vol. 12, no. 3, pp. 426–436, 2021, doi: 10.1007/s13198-021-01087-x.
[5] Z. Wang, X. Yu, N. Feng, and Z. Wang, “An improved collaborative movie recommendation system using computational intelligence,” J. Vis. Lang. Comput., vol. 25, no. 6, pp. 667–675, 2014, doi: 10.1016/j.jvlc.2014.09.011.
[6] G. Adomavicius and A. Tuzhilin, “Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions,” Knowl. Data Eng. IEEE Trans., vol. 17, pp. 734–749, Aug. 2005, doi: 10.1109/TKDE.2005.99.
[7] S. R. S. Reddy, S. Nalluri, S. Kunisetti, S. Ashok, and V. Bachu, “Content-Based Movie Recommendation System Using Genre Correlation: Proceedings of the Second International Conference on SCI 2018, Volume 2,” 2019, pp. 391–397.
[8] J. Son and S. B. Kim, “Content-based filtering for recommendation systems using multiattribute networks,” Expert Syst. Appl., vol. 89, pp. 404–412, 2017, doi: https://doi.org/10.1016/j.eswa.2017.08.008.
[9] Y. Wang, M. Wang, and W. Xu, “A Sentiment-Enhanced Hybrid Recommender System for Movie Recommendation: A Big Data Analytics Framework,” Wirel. Commun. Mob. Comput., vol. 2018, p. 8263704, 2018, doi: 10.1155/2018/8263704.
[10] K. N. Jain, V. Kumar, P. Kumar, and T. Choudhury, “Movie Recommendation System: Hybrid Information Filtering System,” in Intelligent Computing and Information and Communication, 2018, pp. 677–686.
[11] X. Li, W. Jiang, W. Chen, J. Wu, and G. Wang, “HAES: A New Hybrid Approach for Movie Recommendation with Elastic Serendipity,” Aug. 2019, pp. 1503–1512, doi: 10.1145/3357384.3357868.
[12] R. Salakhutdinov and A. Mnih, “Bayesian Probabilistic Matrix Factorization Using Markov Chain Monte Carlo,” in Proceedings of the 25th International Conference on Machine Learning, 2008, pp. 880–887, doi: 10.1145/1390156.1390267.
[13] L. Baltrunas, T. Makcinskas, and F. Ricci, “Group Recommendations with Rank Aggregation and Collaborative Filtering,” in Proceedings of the Fourth ACM Conference on Recommender Systems, 2010, pp. 119–126, doi: 10.1145/1864708.1864733.
[14] T. George and S. Merugu, “A Scalable Collaborative Filtering Framework Based on Co-Clustering,” in Proceedings of the Fifth IEEE International Conference on Data Mining, 2005, pp. 625–628, doi: 10.1109/ICDM.2005.14.