Analisis Komparatif Embedding Semantik Berbasis Large Language Model Pada Sistem Rekomendasi Buku Serendipitous di Perpustakaan Kampus

Rahayu Kartika Sari; Eka Dyar Wahyuni; Amalia Anjani Arifiyanti

doi:10.70340/jirsi.v5i2.443

Authors

Rahayu Kartika Sari Universitas Pembangunan Nasional “Veteran” Jawa Timur
Eka Dyar Wahyuni Universitas Pembangunan Nasional “Veteran” Jawa Timur
Amalia Anjani Arifiyanti Universitas Pembangunan Nasional “Veteran” Jawa Timur

DOI:

https://doi.org/10.70340/jirsi.v5i2.443

Keywords:

Item Embedding, Large Language Model, SASRec, Serendipity, Recommender System, Library

Abstract

The phenomenon of information overload in academic libraries often makes it difficult for users to discover relevant books, which may reduce reading interest. Conventional recommender systems are also prone to filter bubbles and tend to perform poorly under cold-start conditions. This study proposes a sequential recommendation system based on the Self-Attention Based Sequential Recommendation (SASRec) model integrated with five semantic embedding models, namely Word2Vec, BERT Multilingual, OpenAI text-embedding-3-small, Gemini-embedding-001, and Qwen3-Embedding-0.6B, to generate accurate and serendipitous recommendations. In addition, the Serendipity-Oriented Greedy (SOG) re-ranking algorithm is implemented to balance recommendation relevance and serendipity. The data set consists of 14,502 book records and 5,445 user interaction histories after the data cleaning process. Evaluation was conducted under three testing scenarios, namely the all-test set, warm test set, and cold test set, by comparing all model variants before and after the re-ranking process. The results show that the integration of Large Language Model (LLM)-based embeddings consistently improves performance compared to the standard SASRec model and traditional embeddings. Qwen3-Embedding-0.6B achieved the best performance, improving HitRate@10 by up to 282.9% and NDCG@10 by up to 387.8%, while maintaining semantic robustness in cold-start scenarios with an UnSerendipity@K score of 0.613. The implementation of SOG re-ranking reveals a direct trade-off between recommendation accuracy and diversity. Lightweight weighting provides the optimal balance, whereas overly aggressive weighting significantly reduces relevance. The main contribution of this study lies in integrating modern LLM embeddings into a sequential recommendation architecture to improve accuracy and cold-start robustness, while also evaluating the impact of serendipity-oriented re-ranking strategies on balancing recommendation relevance and diversity. Overall, this study demonstrates that modern LLM integration can produce a smarter, more adaptive, and more balanced library recommendation system in terms of both accuracy and serendipity.

Downloads

Download data is not yet available.

References

“Perpusnas RI Mewujudkan Perpustakaan sebagai Pusat Aktivitas Masyarakat - National Library of Indonesia.” Diakses: 12 Desember 2025. [Daring]. Tersedia pada: https://perpusnas.go.id/berita/mewujudkan-perpustakaan-sebagai-pusat-aktivitas-masyarakat

GoodStats, “Punya Perpustakaan Tertinggi di Dunia, Minat Baca di Indonesia Masih Rendah,” GoodStats. Diakses: 8 Desember 2025. [Daring]. Tersedia pada: https://goodstats.id/article/perpus-tertinggi-dunia-namun-minat-baca-rendah-rPe7F

W. Zarman, “Information Overload: Clarifying the Problem,” IJIE Indones. J. Inform. Educ., vol. 5, no. 2, hlm. 79–83, Des 2021, doi: 10.20961/ijie.v5i2.56922.

O. Efe Stanley, “Information Overload: Causes, Symptoms, Consequences and Solutions,” Asian J. Inf. Sci. Technol., vol. 11, no. 2, hlm. 1–6, Nov 2021, doi: 10.51983/ajist-2021.11.2.2887.

L. Shahrzadi, A. Mansouri, M. Alavi, dan A. Shabani, “Causes, consequences, and strategies to deal with information overload: A scoping review,” Int. J. Inf. Manag. Data Insights, vol. 4, no. 2, hlm. 100261, Nov 2024, doi: 10.1016/j.jjimei.2024.100261.

C. C. Aggarwal, Recommender Systems. Cham: Springer International Publishing, 2016. doi: 10.1007/978-3-319-29659-3.

M. Sakinah dan M. Alda, “Sistem Informasi Perpustakaan SMP Negeri 1 Pegajahan Dengan Fitur Notifikasi Untuk Monitoring Peminjaman Buku Berbasis Mobile Dengan Metode Collaborative Filltering,” Teknol. J. Ilm. Sist. Inf., vol. 15, no. 2, hlm. 75–86, Jul 2025, doi: 10.26594/teknologi.v15i2.5733.

M. G. Navsih, A. Muhaimin, dan S. S. M. Wara, “Sistem Rekomendasi Menu Kantin Menggunakan Lifespan-Aware Association Rule Mining Dengan Hybrid Apriori Dan FP-Growth,” Teknol. J. Ilm. Sist. Inf., vol. 16, no. 1, hlm. 13–20, Feb 2026, doi: 10.26594/teknologi.v16i1.6143.

H. Asy’ari, M. Y. Urrochman, dan A. Ro’uf, “Sistem Rekomendasi Kurikulum Adaptif Berbasis Profil Lulusan Menggunakan Metode Hybrid Filtering,” Teknol. J. Ilm. Sist. Inf., vol. 15, no. 2, hlm. 111–119, Sep 2025, doi: 10.26594/teknologi.v15i2.5795.

A. Zainurrohman, “Content-Based Recommender System Using NLP,” MLearning.ai. Diakses: 5 November 2025. [Daring]. Tersedia pada: https://arif-zai-nur-rohman.medium.com/content-based-recommender-system-using-nlp-445ebb777c7a

X. Wu, “Review of collaborative filtering recommendation systems,” Appl. Comput. Eng., vol. 43, no. 1, hlm. 76–82, Feb 2024, doi: 10.54254/2755-2721/43/20230811.

T. M. A. U. Gunathilaka, P. D. Manage, J. Zhang, Y. Li, dan W. Kelly, “Addressing sparse data challenges in recommendation systems: A systematic review of rating estimation using sparse rating data and profile enrichment techniques,” Intell. Syst. Appl., vol. 25, hlm. 200474, Mar 2025, doi: 10.1016/j.iswa.2024.200474.

Q. M. Areeb dkk., “Filter Bubbles in Recommender Systems: Fact or Fallacy -- A Systematic Review,” 2 Juli 2023, arXiv: arXiv:2307.01221. doi: 10.48550/arXiv.2307.01221.

L. Lv, K. Q. Kang, dan G. (Gus) Liu, “Prick ‘filter bubbles’ by enhancing consumers’ novelty-seeking: The role of personalized recommendations of unmentionable products,” Psychol. Mark., vol. 41, no. 10, hlm. 2355–2367, 2024, doi: 10.1002/mar.22057.

L. Chen, Y. Yang, N. Wang, K. Yang, dan Q. Yuan, “How Serendipity Improves User Satisfaction with Recommendations? A Large-Scale User Evaluation,” dalam The World Wide Web Conference, dalam WWW ’19. New York, NY, USA: Association for Computing Machinery, Mei 2019, hlm. 240–250. doi: 10.1145/3308558.3313469.

D. Kotkov, A. Medlar, T. Kask, dan D. Glowacka, “The Dark Matter of Serendipity in Recommender Systems,” dalam Proceedings of the 2024 ACM SIGIR Conference on Human Information Interaction and Retrieval, Sheffield United Kingdom: ACM, Mar 2024, hlm. 108–118. doi: 10.1145/3627508.3638342.

T. F. Boka, Z. Niu, dan R. B. Neupane, “A survey of sequential recommendation systems: Techniques, evaluation, and future directions,” Inf. Syst., vol. 125, hlm. 102427, Nov 2024, doi: 10.1016/j.is.2024.102427.

W.-C. Kang dan J. McAuley, “Self-Attentive Sequential Recommendation,” dalam 2018 IEEE International Conference on Data Mining (ICDM), Nov 2018, hlm. 197–206. doi: 10.1109/ICDM.2018.00035.

D. S. Asudani, N. K. Nagwani, dan P. Singh, “Impact of word embedding models on text analytics in deep learning environment: a review,” Artif. Intell. Rev., hlm. 1–81, Feb 2023, doi: 10.1007/s10462-023-10419-1.

S. Deb dan A. K. Chanda, “Comparative analysis of contextual and context-free embeddings in disaster prediction from Twitter data,” Mach. Learn. Appl., vol. 7, hlm. 100253, Mar 2022, doi: 10.1016/j.mlwa.2022.100253.

A. Boz dkk., “Improving Sequential Recommendations with LLMs,” ACM Trans. Recomm. Syst., hlm. 3711667, Jan 2025, doi: 10.1145/3711667.

J. Hu dkk., “Enhancing Sequential Recommendation via LLM-based Semantic Embedding Learning,” dalam Companion Proceedings of the ACM Web Conference 2024, dalam WWW ’24. New York, NY, USA: Association for Computing Machinery, Mei 2024, hlm. 103–111. doi: 10.1145/3589335.3648307.

Y. He, X. Liu, A. Zhang, Y. Ma, dan T.-S. Chua, “LLM2Rec: Large Language Models Are Powerful Embedding Models for Sequential Recommendation,” 16 Juni 2025, arXiv: arXiv:2506.21579. doi: 10.48550/arXiv.2506.21579.

K. Liu, “When Transformers Meet Recommenders: Integrating Self-Attentive Sequential Recommendation with Fine-Tuned LLMs,” 8 Juli 2025, arXiv: arXiv:2507.05733. doi: 10.48550/arXiv.2507.05733.

D. Kotkov, J. Veijalainen, dan S. Wang, “How does serendipity affect diversity in recommender systems? A serendipity-oriented greedy algorithm,” Computing, vol. 102, no. 2, hlm. 393–411, Feb 2020, doi: 10.1007/s00607-018-0687-5.

G. C. Cawley, “Leave-One-Out Cross-Validation Based Model Selection Criteria for Weighted LS-SVMs,” dalam The 2006 IEEE International Joint Conference on Neural Network Proceedings, Jul 2006, hlm. 1661–1668. doi: 10.1109/IJCNN.2006.246634.

Z. Yuan dkk., “Where to Go Next for Recommender Systems? ID- vs. Modality-based Recommender Models Revisited,” 2 Mei 2023, arXiv: arXiv:2303.13835. doi: 10.48550/arXiv.2303.13835.

S. Karimi, H. A. Rahmani, M. Naghiaei, dan L. Safari, “Provider Fairness and Beyond-Accuracy Trade-offs in Recommender Systems,” 8 September 2023, arXiv: arXiv:2309.04250. doi: 10.48550/arXiv.2309.04250.

L. Pan, W. Pan, M. Wei, H. Yin, dan Z. Ming, “A Survey on Sequential Recommendation,” 17 Desember 2024, arXiv: arXiv:2412.12770. doi: 10.48550/arXiv.2412.12770.

W.-W. Du, T. Udagawa, dan K. Tateno, “Not Just What, But When: Integrating Irregular Intervals to LLM for Sequential Recommendation,” arXiv.org. Diakses: 15 Mei 2026. [Daring]. Tersedia pada: https://arxiv.org/abs/2507.23209v1

B.-C. Chen dan M. Slokom, “LLMDiRec: LLM-Enhanced Intent Diffusion for Sequential Recommendation,” 20 Oktober 2025, arXiv: arXiv:2601.03259. doi: 10.48550/arXiv.2601.03259.

A. Pembek, A. Fatkulin, A. Klenitskiy, dan A. Vasilev, “Let It Go? Not Quite: Addressing Item Cold Start in Sequential Recommendations with Content-Based Initialization,” 25 Juli 2025, arXiv: arXiv:2507.19473. doi: 10.48550/arXiv.2507.19473.

Y. Kim, S. Oh, C. Noh, E. Hong, dan S. Park, “Design of a Serendipity-Incorporated Recommender System,” Electronics, vol. 14, no. 4, hlm. 821, Jan 2025, doi: 10.3390/electronics14040821.

S. Khaitan dan R. Shrivastava, “Developing Fairness, Accuracy, and Serendipity Objective Functions for Recommendation System and Establishing Trade-off through Multi-Objective Evolutionary Optimization,” Inf. Process. Manag., vol. 63, no. 4, hlm. 104604, Jun 2026, doi: 10.1016/j.ipm.2025.104604.

K. Peng, M. Raghavan, E. Pierson, J. Kleinberg, dan N. Garg, “Reconciling the accuracy-diversity trade-off in recommendations,” arXiv.org. Diakses: 15 Mei 2026. [Daring]. Tersedia pada: https://arxiv.org/abs/2307.15142v1