Evaluasi Sistem Rekomendasi Film Berbasis Sinopsis Cold-Start: Perbandingan TF-IDF, Word2Vec, dan FastText: Cold-Start Synopsis-Based Movie Recommendation: Comparing TF-IDF, Word2Vec, and FastText

Khadijah Fahmi Hayati Holle; Ela Ilmatul Hidayah; Renata Amalia Putri; Citra Ayu Putri Ningrum

doi:10.57152/malcom.v6i3.2896

Authors

Khadijah Fahmi Hayati Holle UIN Maulana Malik Ibrahim Malang https://orcid.org/0000-0002-6991-1748
Ela Ilmatul Hidayah Universitas Islam Negeri Maulana Malik Ibrahim
Renata Amalia Putri Universitas Islam Negeri Maulana Malik Ibrahim
Citra Ayu Putri Ningrum Universitas Islam Negeri Maulana Malik Ibrahim

DOI:

https://doi.org/10.57152/malcom.v6i3.2896

Keywords:

Cold-Start, Cosine Similarity, Fasttext, Sistem Rekomendasi, TF-IDF, Word2Vec

Abstract

Penelitian ini bertujuan untuk menentukan representasi teks ringan yang paling efektif dan efisien untuk rekomendasi film cold-start ketika sistem hanya memiliki sinopsis dan belum memiliki riwayat interaksi pengguna. Penelitian membandingkan TF-IDF, Word2Vec, dan FastText yang dikombinasikan dengan cosine similarity pada dataset film IMDb. Dari 10.001 data film, sebanyak 9.924 data valid digunakan setelah penghapusan sinopsis kosong, duplikasi, dan token tidak informatif. Pengujian dilakukan menggunakan 20 skenario kueri cold-start yang terdiri atas sepuluh kueri panjang bernarasi dan sepuluh kueri pendek berbasis kata kunci. Karena dataset tidak menyediakan label relevansi eksplisit dari pengguna, relevansi diukur menggunakan genre-proxy relevance. Evaluasi menggunakan Precision@10, HitRate@10, AP@10, MRR@10, nDCG@10, F1 skenario, serta waktu retrieval rata-rata. Hasil menunjukkan bahwa TF-IDF + cosine similarity memperoleh kinerja terbaik dengan Precision@10 sebesar 0,835, AP@10 sebesar 0,884, MRR@10 sebesar 0,942, nDCG@10 sebesar 0,942, dan F1 skenario sebesar 0,902. Kontribusi utama penelitian ini adalah pembuktian empiris terkontrol bahwa representasi leksikal dapat mengungguli rata-rata word embedding dalam rekomendasi cold-start berbasis sinopsis. TF-IDF lebih unggul karena sinopsis film yang pendek dan kueri tematik yang eksplisit masih mempertahankan kata kunci diskriminatif, sedangkan rata-rata vektor Word2Vec dan FastText cenderung meratakan sinyal kata kunci yang spesifik. Hasil ini menunjukkan bahwa TF-IDF tetap menjadi baseline yang kuat

Downloads

Download data is not yet available.

References

P. Covington, J. Adams, and E. Sargin, “Deep neural networks for YouTube recommendations,” in Proc. 10th ACM Conf. Recommender Systems, 2016, pp. 191–198, doi: 10.1145/2959100.2959190.

S. Zhang, L. Yao, A. Sun, and Y. Tay, “Deep learning based recommender system: A survey and new perspectives,” ACM Comput. Surv., vol. 52, no. 1, pp. 1–38, 2019, doi: 10.1145/3285029.

O. A. S. Ibrahim, E. M. G. Younis, E. A. Mohamed, and A. Abraham, “Revisiting recommender systems: an investigative survey,” Neural Comput. Appl., vol. 37, pp. 2145–2173, 2025, doi: 10.1007/s00521-024-10828-5.

P. Khadka and P. Lamichhane, “Content-based recommendation engine for video streaming platform,” arXiv, 2023, doi: 10.48550/arXiv.2308.08406.

L. Wu, X. He, X. Wang, K. Zhang, and M. Wang, “A survey on accuracy-oriented neural recommendation: From collaborative filtering to information-rich recommendation,” IEEE Trans. Knowl. Data Eng., vol. 35, no. 5, pp. 4425–4445, 2023, doi: 10.1109/TKDE.2022.3145690.

A. H. J. Permana and A. T. Wibowo, “Movie recommendation system based on synopsis using content-based filtering with TF-IDF and cosine similarity,” Int. J. Inf. Commun. Technol., vol. 9, no. 2, pp. 1–14, 2023, doi: 10.21108/ijoict.v9i2.747.

M. Chibb, P. Vashisht, A. Katti, and A. Narang, “Enhancing movie recommendations: A content based approach using TF-IDF weighted Word2Vec and cosine similarity,” in 2024 4th Int. Conf. Technological Advancements in Computational Sciences (ICTACS), 2024, pp. 1449–1454, doi: 10.1109/ICTACS62700.2024.10841087.

A. Febrian and E. D. Permana, “Sistem rekomendasi film menggunakan metode content based filtering dengan algoritma TF-IDF,” Al-Aqlu: J. Mat. Tek. Sains, vol. 4, no. 1, pp. 19–25, 2026, doi: 10.59896/aqlu.v4i1.494.

E. Z. Irabela and Norhikmah, “Implementation of content-based cosine similarity algorithm with TF-IDF and SBERT for movie recommendation,” J. Sistem Cerdas, vol. 9, no. 1, pp. 114–122, 2026, doi: 10.37396/jsc.v9i1.633.

O. Barkan and N. Koenigstein, “Item2Vec: Neural item embedding for collaborative filtering,” in 2016 IEEE 26th Int. Workshop Machine Learning for Signal Processing, 2016, pp. 1–6, doi: 10.1109/MLSP.2016.7738876.

C. González-Santos, M. A. Vega-Rodríguez, J. M. López-Muñoz, I. Martínez-Sarriegui, and C. J. Pérez, “Automatic assignment of microgenres to movies using a word embedding-based approach,” Multimedia Tools Appl., vol. 83, pp. 48719–48735, 2023, doi: 10.1007/s11042-023-17442-y.

A. Fellah, A. Zahaf, and A. Elçi, “Semantic similarity measure using a combination of Word2Vec and WordNet models,” Indones. J. Electr. Eng. Informatics, vol. 12, no. 2, pp. 455–464, 2024, doi: 10.52549/ijeei.v12i2.5114.

P. Bojanowski, E. Grave, A. Joulin, and T. Mikolov, “Enriching word vectors with subword information,” Trans. Assoc. Comput. Linguistics, vol. 5, pp. 135–146, 2017, doi: 10.1162/tacl_a_00051.

X. He, L. Liao, H. Zhang, L. Nie, X. Hu, and T. S. Chua, “Neural collaborative filtering,” in Proc. 26th Int. Conf. World Wide Web, 2017, pp. 173–182, doi: 10.1145/3038912.3052569.

B. Hidasi, A. Karatzoglou, L. Baltrunas, and D. Tikk, “Session-based recommendations with recurrent neural networks,” in Int. Conf. Learning Representations, 2016.

W.-C. Kang and J. McAuley, “Self-attentive sequential recommendation,” in 2018 IEEE Int. Conf. Data Mining, 2018, pp. 197–206, doi: 10.1109/ICDM.2018.00035.

F. Sun, J. Liu, J. Wu, C. Pei, X. Lin, W. Ou, and P. Jiang, “BERT4Rec: Sequential recommendation with bidirectional encoder representations from transformer,” in Proc. 28th ACM Int. Conf. Inf. Knowl. Manage., 2019, pp. 1441–1450, doi: 10.1145/3357384.3357895.

J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186, doi: 10.18653/v1/N19-1423.

N. Reimers and I. Gurevych, “Sentence-BERT: Sentence embeddings using Siamese BERT-networks,” in Proc. EMNLP-IJCNLP, 2019, pp. 3982–3992, doi: 10.18653/v1/D19-1410.

S. Rendle, W. Krichene, L. Zhang, and J. Anderson, “Neural collaborative filtering vs. matrix factorization revisited,” in Proc. 14th ACM Conf. Recommender Systems, 2020, pp. 240–248, doi: 10.1145/3383313.3412488.

R. Cañamares and P. Castells, “On target item sampling in offline recommender system evaluation,” in Proc. 14th ACM Conf. Recommender Systems, 2020, pp. 259–268, doi: 10.1145/3383313.3412259.

E. Zangerle and C. Bauer, “Evaluating recommender systems: Survey and framework,” ACM Comput. Surv., vol. 55, no. 8, pp. 1–38, 2022, doi: 10.1145/3556536.

A. Klimashevskaia, D. Jannach, M. Elahi, and C. Trattner, “A survey on popularity bias in recommender systems,” User Model. User-Adapted Interact., 2024, doi: 10.1007/s11257-024-09406-0.

J. Chen, H. Dong, X. Wang, F. Feng, M. Wang, and X. He, “Bias and debias in recommender system: a survey and future directions,” ACM Trans. Inf. Syst., vol. 41, no. 3, pp. 1–39, 2023, doi: 10.1145/3564284.