Komparasi Transformer T5 dan BART+BERT Untuk Pembangkitan Kata Kunci Otomatis Dari Abstrak Ilmiah
DOI:
https://doi.org/10.30865/json.v7i2.9305Keywords:
Pembangkitan Kata Kunci Otomatis; Model Transformer; T5; BART; BERT; Abstrak IlmiahAbstract
Peningkatan volume publikasi ilmiah menuntut sistem pengindeksan yang efisien, di mana kata kunci memegang peran krusial. Penelitian ini membandingkan efektivitas dua pendekatan berbasis model transformer—T5 dan kombinasi BART+BERT—dalam membangkitkan kata kunci secara otomatis dari abstrak artikel ilmiah. Eksperimen dilakukan menggunakan 100 artikel jurnal multidisipliner dengan evaluasi berbasis ROUGE (kesamaan leksikal) dan Semantic Similarity (kesamaan makna). Hasil menunjukkan bahwa T5 unggul dalam menghasilkan kata kunci eksplisit yang presisi dan cocok untuk pengindeksan teknis, dengan skor Semantic F1 0,5233 terhadap ground truth. Sebaliknya, BART lebih efektif menangkap tema keseluruhan abstrak, menghasilkan kata kunci yang representatif secara semantik dengan Semantic F1 0,7517, sehingga ideal untuk sistem rekomendasi atau klasifikasi topik. Filter BERT terbukti meningkatkan kualitas semantik, sementara Jaccard lebih baik dalam pencocokan leksikal. Temuan ini menekankan pentingnya pemilihan model dan strategi filtering yang disesuaikan dengan tujuan aplikasi sistem pembangkitan kata kunci otomatis.
References
R. Dwi Handayani, Z. Zulkarnaini, W. Wasilah, and R. Herwanto, “Pendampingan Penerapan Intitutional Repository Digital Untuk Peningkatan Current Research Information System,” Dedik. J. Pengabdi. Kpd. Masy., vol. 1, no. 2, pp. 74–84, Aug. 2022, doi: 10.53276/dedikasi.v1i2.31.
L. Corrin, K. Thompson, G. J. Hwang, and J. M. Lodge, “The importance of choosing the right keywords for educational technology publications,” Australas. J. Educ. Technol., vol. 38, no. 2, pp. 1–8, 2022, doi: 10.14742/ajet.8087.
P. Pottier et al., “Title, abstract and keywords: a practical guide to maximize the visibility and impact of academic papers,” Proc. R. Soc. B Biol. Sci., vol. 291, no. 2027, Jul. 2024, doi: 10.1098/rspb.2024.1222.
R. N. Rathi and A. Mustafi, “The importance of Term Weighting in semantic understanding of text: A review of techniques,” Multimed. Tools Appl., vol. 82, no. 7, pp. 9761–9783, 2023, doi: 10.1007/s11042-022-12538-3.
W. Lu, Z. Liu, Y. Huang, Y. Bu, X. Li, and Q. Cheng, “How do authors select keywords? A preliminary study of author keyword selection behavior,” J. Informetr., vol. 14, no. 4, pp. 1–17, 2020, doi: 10.1016/j.joi.2020.101066.
T. Bin Sarwar, N. M. Noor, and M. S. U. Miah, “Evaluating keyphrase extraction algorithms for finding similar news articles using lexical similarity calculation and semantic relatedness measurement by word embedding,” PeerJ Comput. Sci., vol. 8, 2022, doi: 10.7717/peerj-cs.1024.
D. B. Hier et al., “Evaluation of standard and semantically-augmented distance metrics for neurology patients,” BMC Med. Inform. Decis. Mak., vol. 20, no. 1, pp. 1–15, 2020, doi: 10.1186/s12911-020-01217-8.
S. Wang, L. Thompson, and M. Iyyer, “Phrase-BERT: Improved Phrase Embeddings from BERT with an Application to Corpus Exploration,” in Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, Stroudsburg, PA, USA: Association for Computational Linguistics, 2021, pp. 10837–10851. doi: 10.18653/v1/2021.emnlp-main.846.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Jurnal Sistem Komputer dan Informatika (JSON)
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under Creative Commons Attribution 4.0 International License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (Refer to The Effect of Open Access).
