Sistem Rekomendasi Novel Menggunakan Pendekatan Hybrid Berbasis Best Matching 25 (BM25) dan Matrix Factorization

Sattyo Ardicta Pradana; Ilham Saifudin; Lutfi Ali Muharom

doi:10.69693/ijmst.v4i3.12786

Authors

Sattyo Ardicta Pradana Universitas Muhammadiyah Jember
Ilham Saifudin Universitas Muhammadiyah Jember https://orcid.org/0000-0002-2063-4524
Lutfi Ali Muharom Universitas Muhammadiyah Jember

DOI:

https://doi.org/10.69693/ijmst.v4i3.12786

Keywords:

rekomendasi novel, sistem rekomendasi hybrid, BM25, Matrix Factorization, NDCG

Abstract

Pada era digital, jumlah novel yang tersedia di berbagai platform digital berkembang sangat pesat sehingga menyulitkan pembaca dalam memilih judul yang sesuai dengan minat dan preferensinya. Penelitian terdahulu mengenai sistem rekomendasi bacaan umumnya masih berfokus pada buku secara umum dan menggunakan pendekatan tunggal, baik content-based filtering maupun collaborative filtering, yang masing-masing memiliki keterbatasan dalam menangani teks naratif panjang atau permasalahan cold-start. Penelitian ini mengusulkan sistem rekomendasi novel berbasis hybrid yang mengombinasikan metode Best Matching 25 (BM25) untuk mengukur relevansi konten berdasarkan judul, genre, dan sinopsis, serta Matrix Factorization untuk memodelkan preferensi pengguna berdasarkan data rating. Kedua skor digabungkan menggunakan skema hybrid berbobot dengan nilai alpha = 0,6, setelah skor BM25 dinormalisasi menggunakan normalisasi min-max agar sebanding dengan skala rating 1-5 yang dihasilkan oleh Matrix Factorization. Dataset yang digunakan terdiri dari Goodreads-books untuk konten novel dan Goodbooks-10k untuk data rating pengguna, yang disaring sehingga hanya memuat item novel, menghasilkan 8.480 novel, 16.220 pengguna, dan 640.396 rating setelah pre-processing. Kinerja sistem dievaluasi menggunakan NDCG@10 dan Precision@10 pada daftar rekomendasi Top-10. Hasil evaluasi menunjukkan bahwa pendekatan hybrid memperoleh nilai NDCG@10 sebesar 0,943 dan Precision@10 sebesar 0,488, dengan rata-rata runtime 136,46 detik untuk 15.813 pengguna. Pengujian lanjutan terhadap variasi alpha, data split, dan Top-N menunjukkan bahwa Matrix Factorization memberikan kontribusi yang lebih dominan terhadap kualitas pemeringkatan, sedangkan BM25 tetap penting untuk menjaga relevansi berbasis konten. Temuan ini menunjukkan bahwa pendekatan hybrid BM25 dan Matrix Factorization dapat menghasilkan rekomendasi novel yang relevan dan efisien secara komputasi.

References

AL-Ghuribi, S. M., Mohd Noah, S. A., Tiun, S., Mohammed, M. A., & Saat, N. I. Y. (2025). Combining review elements for modelling various multi-criteria collaborative recommendation models. Journal of Big Data, 12(1). https://doi.org/10.1186/s40537-025-01222-6

Ardiansyah, R., Ari Bianto, M., & Saputra, B. D. (2023). Sistem Rekomendasi Buku Perpustakaan Sekolah menggunakan Metode Content-Based Filtering. Jurnal CoSciTech (Computer Science and Information Technology), 4(2), 510–518. https://doi.org/10.37859/coscitech.v4i2.5131

Askari, A., Abolghasemi, A., Pasi, G., Kraaij, W., & Verberne, S. (2023). Injecting the BM25 Score as Text Improves BERT-Based Re-rankers. https://doi.org/10.48550/arXiv.2301.09728

Bojorque, R., & Hurtado, R. (2025). Matrix Factorization-Based Clustering for Sparse Data in Recommender Systems: A Comparative Study. Computation, 13(9). https://doi.org/10.3390/computation13090213

Chaudhari, A., Hitham Seddig, A. A., Sarlan, A., & Raut, R. (2024). A Hybrid Recommendation System: A Review. IEEE Access, 12, 157107–157126. https://doi.org/10.1109/ACCESS.2024.3480693

Fahmi Ilmi, M., & Pandu Adikara, P. (2022). Pencarian Dokumen Skripsi menggunakan BM25 dan Faceted Search berdasarkan Kata Kunci Abstrak (Studi Kasus: Universitas Muhammadiyah Sidoarjo) (Vol. 6, Number 9). https://j-ptiik.ub.ac.id/index.php/j-ptiik/article/view/11531

Kant Yadav, K., Kumar Soni, H., Yadav, G., & Sharma, M. (2023). International Journal of Intelligent Systems And Applications In Engineering Collaborative Filtering Based Hybrid Recommendation System Using Neural Network and Matrix Factorization Techniques. In Original Research Paper International Journal of Intelligent Systems and Applications in Engineering IJISAE (Vol. 2024, Number 8s). https://ijisae.org/index.php/IJISAE/article/view/4307

Kheng, T., Asri, J. S., Wahyu, S., & Yulhendri, Y. (2025). Penerapan Algoritma BM25 dalam Pencarian Lowongan Pekerjaan pada Website Job Portal. Bulletin of Computer Science Research, 5(5), 1029–1038. https://doi.org/10.47065/bulletincsr.v5i5.760

Kong, W.-E., Tai, T.-E., Naveen, P., & Santoso, H. A. (2024). Performance Evaluation on E-Commerce Recommender System based on KNN, SVD, CoClustering and Ensemble Approaches. Journal of Informatics and Web Engineering, 3(3), 63–76. https://doi.org/10.33093/jiwe.2024.3.3.4

Li, X., Lipp, J., Shakir, A., Huang, R., & Li, J. (2024). BMX: Entropy-weighted Similarity and Semantic-enhanced Lexical Search. https://doi.org/10.48550/arXiv.2408.06643

Maramis, G. D. P., Ranti, M. C. P., & Santa, K. (2025). BM25 Algorithm for Improving Academic Website Search Accuracy. Indonesian Journal of Innovation Studies, 27(1). https://doi.org/10.21070/ijins.v27i1.1799

Nesmaoui, R., Louhichi, M., & Lazaar, M. (2023). A Collaborative Filtering Movies Recommendation System based on Graph Neural Network. Procedia Computer Science, 220, 456–461. https://doi.org/10.1016/j.procs.2023.03.058

Peng, S., Xie, X., Zhai, J., Jia, Y., & Gong, Y. (2021). A Page-topic Relevance Algorithm Based on BM25 and Paragraph-Semantic Correlation. Journal of Physics: Conference Series, 1757(1). https://doi.org/10.1088/1742-6596/1757/1/012115

Rajesh, D. B., & Kumar, A. (2025). Collaborative filtering models an experimental and detailed comparative study. Scientific Reports, 15(1). https://doi.org/10.1038/s41598-025-15096-4

Ridhwanullah, D., Kumarahadi, Y. K., & Raharja, B. D. (2024). Content-Based Filtering pada Sistem Rekomendasi Buku Informatika. Jurnal Ilmiah SINUS, 22(2), 57. https://doi.org/10.30646/sinus.v22i2.840

Saifudin, I., & Widiyaningtyas, T. (2024). Systematic Literature Review on Recommender System: Approach, Problem, Evaluation Techniques, Datasets. IEEE Access, 12, 19827–19847. https://doi.org/10.1109/ACCESS.2024.3359274

Saputra, D. M., Angelia, N., & Yusliani, N. (2024). Recommender System for Tourist Destinations in Indonesia Using Matrix Factorization Method. JITSI : Jurnal Ilmiah Teknologi Sistem Informasi, 5(3), 122–127. https://doi.org/10.62527/jitsi.5.3.254

Widiyaningtyas, T., Saifudin, I., Zaeni, I. A. E., Maulana, M. Z. N., & Caesarendra, W. (2025). Memory-based collaborative filtering based on matrix factorization and Gower’s set rank. Journal of King Saud University - Computer and Information Sciences, 37(9). https://doi.org/10.1007/s44443-025-00261-6

Xu, J., Liu, Z., Tan, L., Li, T., Peng, T., & Gong, D. (2025). Local Matrix Factorization With Network Embedding For Recommender Systems. Computing and Informatics, 44, 223–244. https://doi.org/10.31577/cai

Yusfida, F. (2025). A Hybrid Approach for Recommender Systems Based on Alternating Least Squares and CatBoost. Jurnal Teknik Informatika (Jutif), 6(4), 2825–2836. https://doi.org/10.52436/1.jutif.2025.6.4.5002

Zuliuskandar, V. V., Yusa, M., & Purwandari, E. P. (2025). Hybrid Method Using Non-Negative Matrix Factorization And Keyword-Based Filtering For Recommender System In Moocs. Jurnal Teknik Informatika (Jutif), 6(1), 345–358. https://doi.org/10.52436/1.jutif.2025.6.1.3605