GRAPH ATTENTIVE CONVOLUTIONAL NETWORKS DALAM DETEKSI TRANSAKSI ILEGAL PADA JARINGAN BITCOIN UNTUK KEBUTUHAN DIGITAL FORENSIK

GRAPH ATTENTIVE CONVOLUTIONAL NETWORKS DALAM DETEKSI TRANSAKSI ILEGAL PADA JARINGAN BITCOIN UNTUK KEBUTUHAN DIGITAL FORENSIK

• By Lusindah Erdian Nova; Hilman F Pardede;
• Tesis
• 44

Perkembangan teknologi Blockchain telah meningkatkan penggunaan mata
uang kripto, khususnya Bitcoin, dalam berbagai aktivitas transaksi digital. Di balik
berbagai manfaat yang ditawarkan, sifat pseudonim pada jaringan Bitcoin juga
membuka peluang terjadinya aktivitas ilegal, seperti pencucian uang, penipuan, dan
kejahatan siber lainnya. Kondisi tersebut menjadi tantangan dalam proses digital
forensik karena pola transaksi ilegal umumnya tidak dapat dikenali hanya dari
karakteristik setiap transaksi, tetapi juga dipengaruhi oleh hubungan antartransaksi
di dalam jaringan Blockchain.
Penelitian ini bertujuan untuk mengembangkan model deteksi transaksi ilegal
pada jaringan Bitcoin menggunakan Graph Attention Network (GAT), yaitu model
Graph Neural Network yang menerapkan mekanisme attention. Mekanisme
attention pada GAT memungkinkan model memberikan bobot yang berbeda
terhadap setiap transaksi tetangga berdasarkan tingkat relevansinya. Model GAT
kemudian dikombinasikan dengan Class Weight untuk mengatasi
ketidakseimbangan data antara transaksi Licit dan Illicit. Dataset yang digunakan
adalah Elliptic Bitcoin Dataset yang terdiri atas 203.769 transaksi, dengan 42.019
transaksi Licit, 4.545 transaksi Illicit, dan 157.205 transaksi Unknown, serta
234.355 hubungan antartransaksi (edge). Transaksi Unknown tidak digunakan
sebagai target dalam pelatihan dan evaluasi model karena tidak memiliki ground
truth, tetapi tetap dianalisis pada tahap implementasi untuk mengidentifikasi
transaksi yang memiliki pola mendekati kelas Illicit.
Tahapan penelitian meliputi eksplorasi dataset, data preprocessing,
pembentukan graph, pelatihan model, evaluasi performa, serta analisis transaksi
Unknown. Model yang digunakan terdiri atas Support Vector Machine (SVM),
Graph Convolutional Network (GCN), GCN dengan Class Weight, GAT, dan GAT
dengan Class Weight. Evaluasi dilakukan menggunakan metrik Accuracy,
Precision, Recall, F1-score, Confusion Matrix, dan Receiver Operating
Characteristic–Area Under Curve (ROC-AUC).
Hasil penelitian menunjukkan bahwa GAT dengan Class Weight memperoleh
Recall sebesar 90,43%, ROC-AUC sebesar 0,9809, False Negative sebanyak 87
transaksi, dan True Positive sebanyak 822 transaksi. Sementara itu, SVM
x
Program Studi Informatika (S2) FTI Universitas Nusa Mandiri
memperoleh Accuracy tertinggi sebesar 97,36%, tetapi memiliki kemampuan yang
lebih rendah dalam mendeteksi transaksi ilegal dibandingkan model berbasis graph.
Implementasi model GAT + Class Weight terhadap 157.205 transaksi Unknown
menghasilkan 119.100 transaksi (75,76%) yang diprediksi Licit dan 38.105
transaksi (24,24%) yang diprediksi Illicit. Analisis struktur graph juga menemukan
4.027 transaksi Unknown yang memiliki hubungan langsung dengan transaksi
Illicit, dengan 1.942 transaksi Unknown diprediksi sebagai Illicit sekaligus
memiliki hubungan langsung dengan transaksi Illicit. Kelompok transaksi tersebut
dapat digunakan sebagai kandidat prioritas untuk pemeriksaan lebih lanjut, namun
tidak dapat langsung dinyatakan sebagai transaksi ilegal karena tidak memiliki label
aktual.
Berdasarkan hasil penelitian, pendekatan berbasis Graph Neural Network,
khususnya GAT yang dikombinasikan dengan Class Weight, dapat digunakan
untuk mendukung deteksi transaksi ilegal pada jaringan Bitcoin. Selain mampu
mendeteksi transaksi berlabel Illicit, model juga dapat digunakan sebagai alat bantu
untuk menyaring dan memprioritaskan transaksi Unknown yang memiliki indikasi
mencurigakan berdasarkan hasil prediksi model dan hubungan antartransaksi. Hasil
tersebut dapat mendukung proses digital forensik, khususnya dalam membantu
investigator menentukan transaksi yang memerlukan pemeriksaan lebih lanjut.

• Katakunci Bitcoin; blockchain security; cryptocurrency; graph attention network; graph neural network; illicit transaction detection; class imbalance; digital forensics.
• URL http://
• Penerbit ,

REFERENSI

DAFTAR REFERENSI
[1] M. Venturini, D. García-Costa, E. Álvarez-García, F. Grimaldo, and
F. Squazzoni, “Mapping network structures and dynamics of
decentralised cryptocurrencies: The evolution of Bitcoin (2009–
2023),” PLoS One, vol. 20, no. 8 August, Aug. 2025, doi:
10.1371/journal.pone.0329617.
[2] H. F. Atlam, N. Ekuri, M. A. Azad, and H. S. Lallie, “Blockchain
Forensics: A Systematic Literature Review of Techniques,
Applications, Challenges, and Future Directions,” Sep. 01, 2024,
Multidisciplinary Digital Publishing Institute (MDPI). doi:
10.3390/electronics13173568.
[3] Y. Elmougy and L. Liu, “Demystifying Fraudulent Transactions and
Illicit Nodes in the Bitcoin Network for Financial Forensics,” in
Proceedings of the ACM SIGKDD International Conference on
Knowledge Discovery and Data Mining, Association for Computing
Machinery, Aug. 2023, pp. 3979–3990. doi:
10.1145/3580305.3599803.
[4] A. Toha, P. Purwono, and W. Gata, “Model Prediksi Kualitas Udara
dengan Support Vector Machines dengan Optimasi Hyperparameter
GridSearch CV,” Buletin Ilmiah Sarjana Teknik Elektro, vol. 4, no. 1,
pp. 12–21, Apr. 2022, doi: 10.12928/biste.v4i1.6079.
[5] B. Khemani, S. Patil, K. Kotecha, and S. Tanwar, “A review of graph
neural networks: concepts, architectures, techniques, challenges,
datasets, applications, and future directions,” J. Big Data, vol. 11,
no. 1, Dec. 2024, doi: 10.1186/s40537-023-00876-4.
[6] I. Alarab and S. Prakoonwit, “Robust recurrent graph convolutional
network approach based sequential prediction of illicit transactions
in cryptocurrencies,” Multimed. Tools Appl., vol. 83, no. 20, pp.
58449–58464, Jun. 2024, doi: 10.1007/s11042-023-17323-4.
[7] B. Fu, Y. Wang, and T. Feng, “CT-GCN+: a high-performance
cryptocurrency transaction graph convolutional model for phishing
node classification,” Cybersecurity, vol. 7, no. 1, Dec. 2024, doi:
10.1186/s42400-023-00194-5.
[8] A. Asiri and K. Somasundaram, “Graph convolution network for
fraud detection in bitcoin transactions,” Sci. Rep., vol. 15, no. 1, Dec.
2025, doi: 10.1038/s41598-025-95672-w.
152
Program Studi Informatika (S2) FTI Universitas Nusa Mandiri
[9] C. Guo, S. Zhang, P. Zhang, M. Alkubati, and J. Song, “LB-GLAT:
Long-Term Bi-Graph Layer Attention Convolutional Network for
Anti-Money Laundering in Transactional Blockchain,” Mathematics,
vol. 11, no. 18, Sep. 2023, doi: 10.3390/math11183927.
[10] H. Song, Y. Wei, Z. Qu, and W. Wang, “Unveiling Decentralization:
A Comprehensive Review of Technologies, Comparison, Challenges
in Bitcoin, Ethereum, and Solana Blockchain,” Apr. 2024, [Online].
Available: http://arxiv.org/abs/2404.04841
[11] J. Nicholls, A. Kuppa, and N. A. Le-Khac, “The next phase of
identifying illicit activity in Bitcoin,” International Journal of
Network Management, vol. 34, no. 5, Sep. 2024, doi:
10.1002/nem.2259.
[12] O. A. Ekle and W. Eberle, “Anomaly Detection in Dynamic Graphs:
A Comprehensive Survey,” ACM Trans. Knowl. Discov. Data, vol.
18, no. 8, Jul. 2024, doi: 10.1145/3669906.
[13] S. A. Alsaif, “Machine Learning-Based Ransomware Classification
of Bitcoin Transactions,” Applied Computational Intelligence and
Soft Computing, vol. 2023, 2023, doi: 10.1155/2023/6274260.
[14] N. Pocher, M. Zichichi, F. Merizzi, M. Z. Shafiq, and S. Ferretti,
“Detecting anomalous cryptocurrency transactions: An AML/CFT
application of machine learning-based forensics,” Electronic
Markets, vol. 33, no. 1, Dec. 2023, doi: 10.1007/s12525-023-00654-
3.
[15] I. Alarab and S. Prakoonwit, “Graph-Based LSTM for Anti-money
Laundering: Experimenting Temporal Graph Convolutional Network
with Bitcoin Data,” Neural Process. Lett., vol. 55, no. 1, pp. 689–
707, Feb. 2023, doi: 10.1007/s11063-022-10904-8.
[16] M. Li, L. Jia, and X. Q. Su, “Global-local graph attention with cyclic
pseudo-labels for bitcoin anti-money laundering detection,” Sci.
Rep., vol. 15, no. 1, Dec. 2025, doi: 10.1038/s41598-025-08365-9.
[17] L. Garin and V. Gisin, “Machine learning in classifying bitcoin
addresses,” Journal of Finance and Data Science, vol. 9, Nov. 2023,
doi: 10.1016/j.jfds.2023.100109.
[18] V. T. Hoang, H. J. Jeon, E. S. You, Y. Yoon, S. Jung, and O. J. Lee,
“Graph Representation Learning and Its Applications: A Survey,”
Apr. 01, 2023, MDPI. doi: 10.3390/s23084168.
153
Program Studi Informatika (S2) FTI Universitas Nusa Mandiri
[19] F. Guan, T. Zhu, W. Zhou, and K. K. R. Choo, “Graph neural
networks: a survey on the links between privacy and security,” Artif.
Intell. Rev., vol. 57, no. 2, Feb. 2024, doi: 10.1007/s10462-023-
10656-4.
[20] Y. Pan, “Enhancing predictive models for illicit activities in the
Bitcoin transaction network using advanced graph analytical
techniques,” Applied and Computational Engineering, vol. 48, no. 1,
pp. 78–86, Mar. 2024, doi: 10.54254/2755-2721/48/20241181.
[21] H. Adloori, V. Dasanapu, and A. C. Mergu, “Graph Network Models
To Detect Illicit Transactions In Block Chain,” Sep. 2024, [Online].
Available: http://arxiv.org/abs/2410.07150
[22] J. M. H. Pinheiro et al., “The Impact of Feature Scaling in Machine
Learning: Effects on Regression and Classification Tasks,” IEEE
Access, vol. 13, pp. 199903–199931, 2025, doi:
10.1109/ACCESS.2025.3635541.
[23] M. M. Ahsan, M. A. P. Mahmud, P. K. Saha, K. D. Gupta, and Z.
Siddique, “Effect of Data Scaling Methods on Machine Learning
Algorithms and Model Performance,” Technologies (Basel)., vol. 9,
no. 3, Sep. 2021, doi: 10.3390/technologies9030052.
[24] W. Chen, K. Yang, Z. Yu, Y. Shi, and C. L. P. Chen, “A survey on
imbalanced learning: latest research, applications and future
directions,” Artif. Intell. Rev., vol. 57, no. 6, Jun. 2024, doi:
10.1007/s10462-024-10759-6.
[25] L. Xu, H. Zhu, and J. Chen, “Imbalanced graph learning via mixed
entropy minimization,” Sci. Rep., vol. 14, no. 1, Dec. 2024, doi:
10.1038/s41598-024-75999-6.
[26] M. Markwald and E. Demidova, “REFUEL: rule extraction for
imbalanced neural node classification,” Mach. Learn., vol. 113, no.
9, pp. 6227–6246, Sep. 2024, doi: 10.1007/s10994-024-06569-0.
[27] P. Gao, Z. Li, D. Zhou, and L. Zhang, “Reinforced Cost-Sensitive
Graph Network for Detecting Fraud Leaders in Telecom Fraud,”
IEEE Access, vol. 12, pp. 173638–173646, 2024, doi:
10.1109/ACCESS.2024.3448260.
[28] R. Abdulkadirov, P. Lyakhov, and N. Nagornov, “Survey of
Optimization Algorithms in Modern Neural Networks,” Jun. 01,
2023, MDPI. doi: 10.3390/math11112466.
154
Program Studi Informatika (S2) FTI Universitas Nusa Mandiri
[29] J. W. Shim, “Enhancing cross entropy with a linearly adaptive loss
function for optimized classification performance,” Sci. Rep., vol.
14, no. 1, Dec. 2024, doi: 10.1038/s41598-024-78858-6.
[30] R. Connor, A. Dearle, B. Claydon, and L. Vadicamo, “Correlations
of Cross-Entropy Loss in Machine Learning,” Entropy, vol. 26, no.
6, Jun. 2024, doi: 10.3390/e26060491.
[31] M. Owusu-Adjei, J. Ben Hayfron-Acquah, T. Frimpong, and G.
Abdul-Salaam, “Imbalanced class distribution and performance
evaluation metrics: A systematic review of prediction accuracy for
determining model performance in healthcare systems,” PLOS
Digital Health, vol. 2, no. 11 November, Nov. 2023, doi:
10.1371/journal.pdig.0000290.
[32] J. Opitz, “A Closer Look at Classification Evaluation Metrics and a
Critical Reflection of Common Evaluation Practice”, doi:
10.1162/tacl.
[33] D. Chicco and G. Jurman, “The Matthews correlation
coefficient (MCC) should replace the ROC AUC as the standard
metric for assessing binary classification,” BioData Min., vol. 16,
no. 1, Dec. 2023, doi: 10.1186/s13040-023-00322-4.
[34] M. Salmi, D. Atif, D. Oliva, A. Abraham, and S. Ventura, “Handling
imbalanced medical datasets: review of a decade of research,” Artif.
Intell. Rev., vol. 57, no. 10, Oct. 2024, doi: 10.1007/s10462-024-
10884-2.
[35] M. Carpita and S. Golia, “Categorical classifiers in multiclass
classification with imbalanced datasets,” Stat. Anal. Data Min., vol.
16, no. 4, pp. 391–405, Aug. 2023, doi: 10.1002/sam.11624.
[36] M. I. Nugraha and A. Armansyah, “Penutupan Kompetensi Keahlian
SMK dengan Pendekatan Klasifikasi Minat Siswa Menggunakan
Jaringan Syaraf Tiruan,” Jurnal Teknologi Sistem Informasi dan
Aplikasi, vol. 6, no. 3, pp. 269–275, Jul. 2023, doi:
10.32493/jtsi.v6i3.30143.
[37] J. Khatib Sulaiman, D. Safitri, and T. Arita Fitri, “Perbandingan
Algoritma XGBoost dan SVM Dalam Analisis Opini Publik
Pemilihan Presiden 2024,” Indonesian Journal of Computer Science.
[38] R. M. Dhodapkar, E. Li, K. Nwanyanwu, R. Adelman, S.
Krishnaswamy, and J. C. Wang, “Deep learning for quality
assessment of optical coherence tomography angiography images,”
155
Program Studi Informatika (S2) FTI Universitas Nusa Mandiri
Sci. Rep., vol. 12, no. 1, Dec. 2022, doi: 10.1038/s41598-022-17709-
8.
[39] D. Chicco, M. J. Warrens, and G. Jurman, “The Matthews
Correlation Coefficient (MCC) is More Informative Than Cohen’s
Kappa and Brier Score in Binary Classification Assessment,” IEEE
Access, vol. 9, pp. 78368–78381, 2021, doi:
10.1109/ACCESS.2021.3084050.
[40] P. Christen, D. J. Hand, and N. Kirielle, “A Review of the FMeasure:
Its History, Properties, Criticism, and Alternatives,” Mar.
31, 2024, Association for Computing Machinery. doi:
10.1145/3606367.
[41] S. - and A. -, “Klasifikasi Status Kemiskinan Menggunakan
Algoritma Random Forest,” EIGEN MATHEMATICS JOURNAL, pp.
37–44, Jun. 2022, doi: 10.29303/emj.v5i1.133.
[42] S. Jung, T. Dagobert, J. M. Morel, and G. Facciolo, “A Review of t-
SNE,” Image Processing On Line, vol. 14, pp. 250–270, 2024, doi:
10.5201/ipol.2024.528.
[43] S. Ouyang, Q. Bai, H. Feng, and B. Hu, “Bitcoin Money Laundering
Detection via Subgraph Contrastive Learning,” Entropy, vol. 26, no.
3, Mar. 2024, doi: 10.3390/e26030211.
[44] Z. Li and E. He, “Graph Neural Network-Based Bitcoin Transaction
Tracking Model,” IEEE Access, vol. 11, pp. 62109–62120, 2023,
doi: 10.1109/ACCESS.2023.3288026.
[45] D. Cheng, Y. Ye, S. Xiang, Z. Ma, Y. Zhang, and C. Jiang, “Anti-
Money Laundering by Group-Aware Deep Graph Learning,” IEEE
Trans. Knowl. Data Eng., vol. 35, no. 12, pp. 12444–12457, Dec.
2023, doi: 10.1109/TKDE.2023.3272396.
[46] F. Wan and P. Li, “A Novel Money Laundering Prediction Model
Based on a Dynamic Graph Convolutional Neural Network and
Long Short-Term Memory,” Symmetry (Basel)., vol. 16, no. 3, Mar.
2024, doi: 10.3390/sym16030378.
[47] W. W. Lo, G. K. Kulatilleke, M. Sarhan, S. Layeghy, and M.
Portmann, “Inspection-L: Self-Supervised GNN Node Embeddings
for Money Laundering Detection in Bitcoin,” Oct. 2022, [Online].
Available: http://arxiv.org/abs/2203.10465
156
Program Studi Informatika (S2) FTI Universitas Nusa Mandiri
[48] J. Liu et al., “Graph Embedding-Based Money Laundering Detection
for Ethereum,” Electronics (Switzerland), vol. 12, no. 14, Jul. 2023,
doi: 10.3390/electronics12143180.
[49] Q. Wang, W. T. Tsai, and B. Du, “RMGANets: reinforcement
learning-enhanced multi-relational attention graph-aware network
for anti-money laundering detection,” Complex and Intelligent
Systems, vol. 11, no. 1, Jan. 2025, doi: 10.1007/s40747-024-01615-9.
[50] Z. Lin et al., “Detecting illicit transactions in bitcoin: a wavelettemporal
graph transformer approach for anti-money laundering,”
Sci. Rep., vol. 16, no. 1, Dec. 2026, doi: 10.1038/s41598-025-23901-
3.
[51] A. G. Vrahatis, K. Lazaros, and S. Kotsiantis, “Graph Attention
Networks: A Comprehensive Review of Methods and Applications,”
Sep. 01, 2024, Multidisciplinary Digital Publishing Institute
(MDPI). doi: 10.3390/fi16090318.
[52] K. Wang, Y. Ding, and S. C. Han, “Graph neural networks for text
classification: a survey,” Artif. Intell. Rev., vol. 57, no. 8, Aug. 2024,
doi: 10.1007/s10462-024-10808-0.
[53] K. H. Almotairi, “Towards partial autonomy of operation and
maintenance of unreliable equipment,” Heliyon, vol. 9, no. 9, Sep.
2023, doi: 10.1016/j.heliyon.2023.e20055.
[54] M. C. Hinojosa Lee, J. Braet, and J. Springael, “Performance
Metrics for Multilabel Emotion Classification: Comparing Micro,
Macro, and Weighted F1-Scores,” Applied Sciences (Switzerland),
vol. 14, no. 21, Nov. 2024, doi: 10.3390/app14219863.