Pembangunan Multi-criteria Recommender System Dengan Metode Collaborative Filtering Dalam Studi Kasus Rekomendasi Produk Kecantikan
DOI:
https://doi.org/10.34818/eoe.v8i6.16986Abstract
Abstrak Salah satu metode dasar metode yang dapat diterapkan dalam pembangunan recommender systems adalah Collaborative Filtering. Untuk menaikkan kualitas hasil dari recommender system, dapat diterapkan bentuk multicriteria dalam sistem rating-nya, yang sebelumnya hanya berbentuk single-criteria. Penerapan multi-criteria yang dimaksud yakni melibatkan aspek tambahan selain nilai rating secara overall, dalam studi kasus ini terdapat penilaian terhadap packaging, penilaian pembelian kembali, dan penilaian kelayakan harga sebagai aspek-aspek tersebut. Dalam penelitian ini, digunakan beberapa pilihan konfigurasi, yakni pendekatan user-based collaborative filtering, perhitungan similarity menggunakan euclidean distance dan prediksi rating menggunakan average dari neighborhood. Untuk menguji kualitas dari penelitian ini, dilakukan 2 jenis eksperimen yang keduanya diukur menggu-nakan MAE. Uji yang pertama untuk mengetahui jumlah neighborhood terbaik, hasil uji menunjukkan bahwa kualitas recommender system terbaik didapatkan dengan konfigurasi jumlah neighborhood sejum-lah 80 dengan MAE sebesar 2.4197. Sedangkan pengujian yang kedua adalah membandingkan kualitas sistem single-criteria dengan sistem multi-criteria. Sistem single-criteria memperoleh MAE sebesar 3.5906 dan multi-criteria memperoleh MAE sebesar 2.4197 Kata kunci : sistem rekomendasi, collaborative filtering, multi-criteria Abstract One of the basic methods that can be applied in recommender systems development is Collaborative Filte-ring. To improve the quality of the results of the recommender system, a multi-criteria form can be applied in the rating system, which was previously only in the form of single criteria. The application of multi-criteria in question that involves additional aspects other than the overall rating value, in this case study there are packaging rating, repurchase rating, and the appraisal of the price as these aspects. In this study, several configuration options were used, namely the userbased collaborative filtering approach, similarity calculation using Euclidean distance and rating prediction using the neighborhood average. To measure the quality of this study, 2 types of experiments were conducted, both of which were measured using MAE. The first test was to determine the number of the best neighborhood, the test results showed that the best quality of recommender system was obtained by configuring the number of neighborhood number 80 with an MAE of 2.4197. While the second test is to compare the quality of the single-criteria system with the multicriteria system. The single-criteria system obtained an MAE of 3.5906 and the multi-criteria system obtained an MAE of 2.4197 Keywords: recommender systems, collaborative filtering, multi-criteriaReferences
[1] G. Adomavicius and Y. Kwon. New recommendation techniques for multicriteria rating systems. IEEE
Intelligent Systems, 22(3):48–55, 2007.
[2] C. C. Aggarwal. Recommender Systems: the textbook. Springer, 2018.
[3] D. K. Bokde, S. Girase, and D. Mukhopadhyay. An approach to a university recommendation by multicriteria collaborative filtering and dimensionality reduction techniques. In 2015 IEEE International Symposium on Nanoelectronic and Information Systems, pages 231–236. IEEE, 2015.
[4] J. L. Herlocker, J. A. Konstan, L. G. Terveen, and J. T. Riedl. Evaluating collaborative filtering recommender
systems. ACM Transactions on Information Systems (TOIS), 22(1):5–53, 2004.
[5] D. Jannach, Z. Karakaya, and F. Gedikli. Accuracy improvements for multi-criteria recommender systems.
In Proceedings of the 13th ACM conference on electronic commerce, pages 674–689. ACM, 2012.
[6] J. Lu, D. Wu, M. Mao, W. Wang, and G. Zhang. Recommender system application developments: a survey.
Decision Support Systems, 74:12–32, 2015.
[7] N. Manouselis and C. Costopoulou. Experimental analysis of design choices in multiattribute utility collaborative filtering. International Journal of Pattern Recognition and Artificial Intelligence, 21(02):311–331,
2007.
[8] C. Musto, M. de Gemmis, G. Semeraro, and P. Lops. A multi-criteria recommender system exploiting aspectbased sentiment analysis of users’ reviews. In Proceedings of the eleventh ACM conference on recommender
systems, pages 321–325. ACM, 2017.
[9] P. Resnick and H. R. Varian. Recommender systems. Communications of the ACM, 40(3):56–59, 1997.
[10] F. Ricci, L. Rokach, and B. Shapira. Recommender systems handbook. Springer, 2015.
[11] Y. Zheng. Criteria chains: a novel multi-criteria recommendation approach. In Proceedings of the 22nd
International Conference on Intelligent User Interfaces, pages 29–33. ACM, 2017.
