cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Asa Hari Wibowo
Contact Email
asa.hari@uho.ac.id
Phone
+6285299311848
Journal Mail Official
semantik.informatika@uho.ac.id
Editorial Address
Informatics Engineering Department of Halu Oleo University, Engineering Faculty Building 3rd Floor H.E.A. Mokodompit Street, Bumi Tridharma Green Campus, Halu Oleo University
Location
Kota kendari,
Sulawesi tenggara
INDONESIA
semanTIK
Published by Universitas Halu Oleo
ISSN : 24601446     EISSN : 25028928     DOI : http://dx.doi.org/10.55679/semantik.v8i1
Core Subject : Science, Engineering,
Computer Science & IT Control & Systems Engineering
Jurnal semanTIK is a is one of the media publication of research results in the field of information technology. semanTIK is published Biannually, January-June and July-December and provide scientific publication medium for researchers, engineers, practitioners, academicians, and observers in the field related to semanTIK Focus & Scope. This journal accepts original papers, review articles, case studies, and short communications. The articles published are peer-reviewed by one or two reviewers and cover various Informatics subjects related to the field journal include Software Engineering, Computer Networking, Intelligent Systems, Information Systems, Robotics, Computational Science, Geographic Information Systems, and all topics which related to informatics. The targets in publishing this journal are Lecturers, Students, and Researchers in IT. The paper published in this journal implies that the work described has not been, and will not be published elsewhere, except as part of a lecture, review.
Arjuna Subject : Ilmu Komputer (semua kategori) - Aplikasi Ilmu Komputer
Articles 64 Documents
 3   4   5   6   7 
Implementasi Sistem Parkir Cerdas Berbasis IoT dan QR Code dengan OCR untuk Segmentasi Plat Nomor Kendaraan Muhammad Dio Alfajri; Suroso; Irma Salamah
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.202

Abstract

Fokus penelitian ini adalah membuat sistem parkir cerdas otomatis (IoT) yang menggunakan teknologi QR Code dan kecerdasan buatan (AI) untuk mengidentifikasi kendaraan dengan lebih cepat dan akurat. Meningkatnya jumlah kendaraan di kawasan perkotaan menyebabkan permasalahan parkir yang semakin kompleks, terutama pada sistem konvensional yang masih bergantung pada pencatatan manual. Kondisi ini menimbulkan antrean panjang, kesalahan pendataan, dan rendahnya efisiensi pengelolaan. Oleh karena itu, dibutuhkan solusi parkir modern yang cerdas, cepat, dan akurat. Penelitian ini bertujuan untuk merancang dan mengimplementasikan sistem parkir cerdas otomatis berbasis Internet of Things (IoT) dengan integrasi QR Code sebagai identifikasi kendaraan dan Optical Character Recognition (OCR) berbasis kecerdasan buatan untuk pengenalan plat nomor. Metode penelitian menggunakan pendekatan eksperimental, mencakup perancangan perangkat keras dan perangkat lunak, integrasi sensor barcode GM66, kamera untuk akuisisi citra, mikrokontroler ESP32 sebagai pusat kendali, serta motor servo TD8120MG sebagai aktuator portal parkir. Sistem juga dilengkapi dengan algoritma segmentasi citra untuk mengekstraksi karakter plat nomor dan integrasi Telegram API untuk notifikasi real-time kepada pengguna. Hasil implementasi menunjukkan bahwa sistem mampu meningkatkan efisiensi waktu tunggu kendaraan, dengan akurasi identifikasi rata-rata sebesar 90%. Integrasi teknologi IoT dan AI menjadikan sistem ini sebagai solusi inovatif yang layak diterapkan dalam pengelolaan parkir modern di berbagai fasilitas publik seperti kampus, perkantoran, dan pusat layanan umum.  The focus of this research is to create an automatic smart parking system (IoT) that uses QR Code technology and artificial intelligence (AI) to identify vehicles more quickly and accurately. The increasing number of vehicles in urban areas has led to increasingly complex parking problems, especially in conventional systems that still rely on manual recording. This situation causes long queues, data entry errors, and low management efficiency. Therefore, a modern parking solution that is smart, fast, and accurate is needed. This study aims to design and implement an Internet of Things (IoT)-based smart automatic parking system with QR Code integration for vehicle identification and artificial intelligence-based Optical Character Recognition (OCR) for license plate recognition. The research method uses an experimental approach, including hardware and software design, integration of GM66 barcode sensors, cameras for image acquisition, ESP32 microcontrollers as control centers, and TD8120MG servo motors as parking portal actuators The implementation results show that the system is capable of reducing vehicle waiting times, with an average identification accuracy of 90%. The integration of IoT and AI technologies makes this system an innovative solution that is suitable for use in modern parking management in various public facilities such as campuses, offices, and public service centers
Analisis Keamanan Sistem Informasi Website Kampus Menggunakan Metode Penetration Test I Kadek Ryan Jody Prayoga; Putu Wida Gunawan; I Nyoman Bernadus
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.212

Abstract

SIISTA (Single Integrated Information System Undhira) adalah salah satu sistem informasi pengelolaan data dan penyedia layanan kampus yang belum diuji keamanan dan kerentanan sistemnya dari serangan siber, sehingga dibutuhkan upaya untuk memastikan dan menganalisis keamanan sistem dari potensi ancaman serangan siber. Penelitian ini bertujuan untuk menganalisis dan menguji sistem informasi SIISTA dengan menggunakan metode penetration testing dengan melakukan XSS injection dan percobaan brute force pada sistem. Metode penetration testing melalui lima tahapan utama yaitu perencanaan, pengumpulan informasi, analisis kerentanan, eksploitasi, pemeliharaan akses dan analisis hasil. Tools yang digunakan dalam proses ini antara lain Nmap, burp suite, OWASP ZAP, dan metasploit framework. Hasil pengujian menunjukkan adanya kerentanan pada sistem, yaitu kerentanan reflected XSS persistent-effect, reflected XSS, dan penggunaan JS library versi lama yang rentan terhadap eksploitasi dan serangan siber. Selain itu, hasil eksploitasi tidak menemukan akses tidak sah ke dalam sistem secara langsung, namun potensi dari serangan terhadap input pengguna tetap tinggi dan bisa dimanfaatkan untuk phishing atau pencurian data. Berdasarkan hasil pengujian, kesimpulannya adalah sistem informasi SIISTA masih memiliki kerentanan yang dapat dimanfaatkan oleh pihak yang tidak bertanggung jawab, meskipun tidak ada akses langsung yang berhasil diperoleh kedalam sistem, kerentanan sanitasi input atau XSS pada sistem dapat dimanfaatkan secara berulang. SIISTA (Single Integrated Information System Undhira) is a campus data management and service information system that has not yet undergone security and vulnerability testing against potential cyberattacks. Therefore, it is necessary to evaluate and analyze the system’s security to identify possible threats. This study aims to assess the security of the SIISTA information system using the penetration testing method by performing XSS injections and brute-force attempts. The penetration testing process follows five main phases: planning, information gathering, vulnerability analysis, exploitation, access maintenance, and reporting. The tools used in this process include Nmap, Burp Suite, OWASP ZAP, and the Metasploit Framework. The results indicate the presence of several vulnerabilities in the system, such as reflected XSS with persistent effect, standard reflected XSS, and the use of outdated JavaScript libraries that are susceptible to exploitation and cyberattacks. While no unauthorized access to the system was achieved during exploitation, the high potential for attacks targeting user input could lead to phishing or data theft. Based on the findings, it can be concluded that the SIISTA information system still contains vulnerabilities that may be exploited by malicious parties. Although direct access to the system was not obtained, input sanitization flaws such as XSS can be repeatedly exploited.
Klasifikasi Penyakit Osteoarthritis Pada Citra Tulang Lutut Menggunakan Metode Sobel dan Learning Vector Quantization Suhendro Busono; Auliyaur Robbani
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.219

Abstract

Osteoarthritis merupakan penyakit degeneratif yang menyerang tulang rawan, ligamen, dan tulang sehingga menimbulkan rasa nyeri dan kaku. Osteoarthritis merupakan penyakit degeneratif pada tulang seperti sendi yang terdapat pada tangan, lutut, leher bahkan pinggang. Osteoarthritis dapat menimbulkan rasa nyeri, kaku, bahkan deformitas pada sendi. Penyakit Osteoarthritis memiliki 5 kategori dimana penggolongan ini berdasarkan tingkat keparahan yang diderita oleh pasien. Tujuan dari penelitian ini dibagi menjadi dua bagian, tujuan pertama adalah untuk memperkuat hasil pengamatan yang telah dilakukan oleh dokter spesialis ortopedi melalui pendekatan ilmu komputer dan ilmu pengolahan citra digital. Tujuan kedua adalah penelitian ini bermanfaat untuk mewujudkan SDG (Sustainable Development Goal) khususnya kehidupan yang sehat dan sejahtera. Penelitian ini menggunakan metode sobel untuk mendapatkan pola citra tulang lutut dan metode LVQ (Learning Vector Quantization) untuk mendapatkan klasifikasi tingkat penyakit Osteoarthritis yang diderita oleh pasien. Metode ini memiliki beberapa tahapan. Tahapan pertama adalah pengumpulan dan klasifikasi data. Tahapan kedua adalah preprocessing data. Tahapan ketiga adalah implementasi dan pengujian. Tahapan terakhir adalah analisis dan evaluasi. Penelitian ini mendapatkan kinerja terbaik dengan nilai akurasi sebesar 69,94% nilai Specitivity sebesar 72.64% dan nilai Sensitivity sebesar 75.51% dimana parameter yang digunakan adalah jumlah iterasi sebanyak 100 iterasi, jumlah hidden layer sebanyak sepuluh layer dan nilai learning rate sebesar 0,1. Osteoarthritis is a degenerative disease that involves the cartilage, ligament, and bones causing pain and stiffness. Osteoarthritis is a degenerative disease of bones such as joints found in the hands, knees, and neck and even waist. Osteoarthritis can cause pain, stiffness, and even deformity of the joints. Osteoarthritis disease has 5 categories where this classification is based on the severity suffered by the patient. The purpose of this research is divided into two parts, the first goal is to strengthen the observations made by orthopedic specialist doctors through the approach of computer science and digital image processing science. The second goal is this research is useful for realizing SDG (Sustainable Development Goal), especially a healthy and prosperous life. This research uses the sobel method to get the pattern of the knee bone image and the LVQ (Learning Vector Quantization) method to get the classification of the level of Osteoarthritis disease suffered by the patient. This method has several level. The first lavel is data collection and classification. The second level is data preprocessing. The third level is implementation and testing. The last level is analysis and evaluation. This research gets the best performance with an accuracy value of 69.94%, a Specitivity value of 72.64% and a Sensitivity value of 75.51% where the parameters used are the number of iterations are 100 iterations, the number of hidden layers are ten layers and the learning rate value is 0.1.
Penerapan Algoritma K-Means Clustering untuk Pengelompokan Zona Nilai Tanah Berbasis WebGIS Arsila Duaulu; Glenn David Paulus Maramis; Kristofel Santa
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.220

Abstract

Akses terhadap data spasial yang akurat dan interaktif penting untuk administrasi pertanahan, khususnya pengelolaan Zona Nilai Tanah (ZNT). Di Kota Tomohon, penyebaran ZNT masih bersifat konvensional melalui peta cetak atau file statis, sehingga rawan keterlambatan pembaruan dan kesalahan interpretasi. Penelitian ini mengembangkan aplikasi WebGIS dengan algoritma K-Means Clustering untuk mengklasifikasikan ZNT di beberapa kecamatan Tomohon Barat, Selatan, dan Tengah, serta memanfaatkan CesiumJS untuk visualisasi 3D interaktif. Pengembangan menggunakan metodologi Waterfall melalui tahapan analisis, desain, implementasi, pengujian, dan pemeliharaan. Aplikasi memungkinkan pengguna melihat peta interaktif, mengunduh data, memfilter nilai, dan mengelola shapefile secara dinamis. Pengujian black-box memastikan semua fitur berfungsi dengan baik, sedangkan validasi spasial melalui confusion matrix menghasilkan akurasi 82,5% (K=5). Sistem ini meningkatkan transparansi, aksesibilitas, dan efisiensi layanan pertanahan serta mendukung transformasi digital Kantor Pertanahan Tomohon. Access to accurate and interactive spatial data is essential for land administration, particularly for managing Land Value Zones (ZNT). In Tomohon City, ZNT dissemination is still conventional through printed maps or static files, causing delayed updates and potential interpretation errors. This study develops a WebGIS application implementing K-Means Clustering to classify ZNT in several sub-districts of Tomohon Barat, Selatan, and Tengah, and integrates CesiumJS for interactive 3D visualization. The system was built using the Waterfall methodology through analysis, design, implementation, testing, and maintenance phases. The application allows users to view interactive maps, download data, apply value-based filters, and manage shapefiles dynamically. Black-box testing confirmed that all features worked properly, while spatial validation using a confusion matrix achieved 82.5% accuracy (K=5). The system improves transparency, accessibility, and efficiency of land information services and supports the digital transformation of the Tomohon Land Office.
Penerapan Metode FMEA dalam Penilaian Risiko Sistem Pemerintahan Berbasis Elektronik Holis Hermansyah; Mustarum Musaruddin; Hasmina Tari Mokui; Muh. Nadzirin Anshari Nur; Abdul Kadir
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.230

Abstract

  Sistem Pemerintahan Berbasis Elektronik (SPBE) adalah penyelenggaraan Pemerintahan yang memanfaatkan Teknologi Informasi dan Komunikasi. Penerapan Teknologi Informasi pada SPBE ini ternyata menimbulkan banyak risiko. Mulai dari risiko sistem hingga keamanan informasi. Risiko sistem dapat berupa kegagalan perangkat keras, kegagalan perangkat lunak hingga pemadaman listrik. Adapun risiko keamanan dapat berupa serangan siber, kebocoran data dan lain sebagainya. Oleh karena itu diperlukan adanya manajemen risiko SPBE berbasis ISO 31000 dan Permen PANRB No. 5 Tahun 2020 guna meminimalisir terjadinya risiko tersebut. Penelitian ini bertujuan untuk menganalisis risiko SPBE pada Dinas Komunikasi dan Informatika Kabupaten Konawe. Metode analisis risiko yang digunakan adalah Failure Mode and Effect Analysis. Hasil identifikasi risiko didapatkan 11 risiko potensial. Dari 11 risiko tersebut kemudian dianalisis menggunakan FMEA dan didapatkan sebanyak 7 risiko menjadi prioritas penanganan karena memiliki dampak yang signifikan. The Electronic Government System refers to the implementation of governance by utilizing Information and Communication Technology. However, the application of Information Technology (IT) within E-Government introduces various risks, ranging from system-related issues to information security concerns. Therefore, E-Government risk management, based on ISO 31000 and the Minister of Administrative and Bureaucratic Reform Regulation (Permen PANRB) No. 5 of 2020, is necessary to mitigate these risks. This study aims to analyze SPBE risks within the Department of Communication and Informatics of Konawe Regency. The risk analysis methods used are Failure Mode and Effect Analysis (FMEA). The risk identification process revealed 11 potential risks. Of these 11 risks, further analysis using FMEA determined that seven were prioritized for handling due to their significant impact.  
Evaluation of Employee Performance at PS Press Rental Start-Up in Semarang Using AHP Method Aryasatya Abdie Maheswara; Iwan Setiawan
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.231

Abstract

Employee performance evaluation is essential for maintaining service quality and operational efficiency, especially in the entertainment industry. However, conventional evaluation methods often suffer from subjectivity and inconsistency. This study aims to implement the Analytical Hierarchy Process (AHP) as a decision support method to enhance objectivity and accuracy in employee performance appraisal at PS Press Start Id Rental Semarang. The AHP approach structures performance evaluation into a hierarchy of criteria productivity, competence, attendance, and service and applies pairwise comparisons to assign weights to each. The study involved designing a web-based evaluation system incorporating the AHP method. Results show that the AHP method successfully minimized subjectivity by ensuring a consistency ratio below the accepted threshold of 0.1. Employee performance was quantified and ranked, with Employee C identified as the top performer. Compared to the previous manual method, evaluation accuracy improved by 18% and decision-making time was reduced by 25%. Employee performance scores ranged from 0.62 to 0.88, with Employee C ranked highest. The system delivered accurate, transparent, and data-driven results, enhancing HR decisions related to rewards, promotions, and development planning. This research confirms AHP’s effectiveness in employee evaluation and offers a scalable model for similar organizations. These traditional approaches often lacked standardized weighting, leading to inconsistent scores and subjective variations of up to ±15% between evaluators. With the AHP-based system, the consistency ratio was reduced to 0.06 well below the accepted 0.1 threshold ensuring higher reliability. The system provided accurate, transparent, and data-driven results, improving HR decision-making processes related to rewards, promotions, and development planning. This research confirms AHP’s effectiveness in employee evaluation and offers a scalable model for similar organizations.
Feature Oriented Competitor Analysis pada Warehouse Management System di Industri Razzan Carveyna; Wahyu Andhyka Kusuma
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.232

Abstract

Warehouse Management System (WMS) memiliki peran penting dalam mendukung efisiensi manajemen stok gudang, namun analisis fitur kompetitor secara sistematis masih jarang dilakukan. Penelitian ini bertujuan untuk menganalisis fitur utama WMS dan memberikan rekomendasi pengembangan aplikasi ScanMP sebagai sistem manajemen stok dengan integrasi barcode scanner dan otomatisasi laporan. Metode yang digunakan adalah kualitatif deskriptif dengan pendekatan Feature Oriented Competitor Analysis (FOCA) berbasis App Store dan Large Language Model (LLM), melalui analisis metadata dari lima aplikasi kompetitor yang diperoleh dengan web scraping pada Google Play Store. Hasil penelitian mengidentifikasi lima fitur utama yang paling umum, yaitu Stock Tracking, Barcode Scanning, Report Export, Expense Tracking, dan User Access Control, yang dipetakan dalam bentuk hierarki fitur. Selain itu, penelitian ini juga menyoroti relevansi fitur tersebut terhadap kebutuhan industri yang dinamis. Kesimpulan dari penelitian ini menunjukkan bahwa pendekatan FOCA berbasis App Store dan LLM mampu memberikan wawasan strategis bagi pengembangan ScanMP agar lebih adaptif dan kompetitif dalam memenuhi kebutuhan industri logistik. Warehouse Management Systems (WMS) play a crucial role in enhancing warehouse stock efficiency, yet systematic competitor feature analysis remains limited. This study aims to analyze the core features of WMS applications and provide recommendations for developing ScanMP, a stock management system with barcode scanning and automated reporting capabilities. A descriptive qualitative method was applied using the Feature Oriented Competitor Analysis (FOCA) approach supported by a Large Language Model (LLM), by analyzing metadata from five competitor applications collected through web scraping of the Google Play Store. The results identified five core features—Stock Tracking, Barcode Scanning, Report Export, Expense Tracking, and User Access Control—which were mapped into a hierarchical structure. In addition, this study emphasizes the relevance of these features to the evolving demands of the industry. The findings conclude that the FOCA approach with LLM provides strategic insights to support the development of ScanMP as a more adaptive and competitive WMS for industrial needs.
Optimasi Model Deep Learning untuk Klasifikasi Stunting berdasarkan Data Antropometri dan Status Imunisasi mardiawati mardiawati; Alders Paliling; Nurul Mutmainnah
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.233

Abstract

Stunting adalah masalah gizi jangka panjang yang memengaruhi pertumbuhan dan perkembangan anak. Klasifikasi status stunting sangat penting untuk mencegah dampak negatif jangka panjang terhadap kualitas hidup anak. Penelitian ini bertujuan untuk mengembangkan dan mengoptimalkan model Deep Learning dengan arsitektur Multilayer Perceptron (MLP), yaitu jenis jaringan syaraf tiruan berlapis yang mampu mempelajari pola non-linear secara efektif, dalam klasifikasi status stunting berdasarkan data Antropometri dan status imunisasi. Dataset yang digunakan terdiri dari 78 data balita, dengan distribusi seimbang antara stunting dan normal. Data yang digunakan masih tergolong kecil karena hanya mengambil sampel pada satu puskesmas. Model dilatih dengan algoritma Adam, menggunakan proses normalisasi data dan teknik early stopping untuk mencegah overfitting. Hasil evaluasi model menunjukkan kinerja yang sangat baik, dengan akurasi mencapai 87.5%, precision 1.00, recall 0.78, dan F1 Score 0.875. Temuan ini menunjukkan bahwa pendekatan berbasis MLP dapat menjadi alternatif yang efektif dalam mendukung proses klasifikasi status stunting secara otomatis dan akurat. Stunting is a long-term nutritional problem that affects children's growth and development. Early detection of stunting status is very important to prevent long-term negative impacts on children's quality of life. This research aims to develop and optimise Deep Learning models with Multilayer Perceptron (MLP) architecture in the classification of stunting status based on Anthropometric data and immunisation status. The dataset used consists of 78 toddler data, with a balanced distribution between stunting and normal. The model was trained with Adam's algorithm, using data normalisation process and early stopping technique to prevent overfitting. The model evaluation results showed excellent performance, with accuracy reaching 87.5%, precision 1.00, recall 0.78, and F1 Score 0.875. These findings suggest that MLP-based approaches can be an effective alternative in supporting the automatic and accurate classification of stunting status.
Integrating Off-Chain and On-Chain Mechanisms to Enhance Medical Data Management Using Blockchain Technology herman herman; Rinday Zildjiani Salji; Herman Yuliansyah
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.235

Abstract

Manajemen data medis yang aman dan andal merupakan tantangan utama dalam sistem perawatan kesehatan digital, di mana pertumbuhan eksponensial rekam medis elektronik meningkatkan permintaan akan solusi penyimpanan yang scalable dan efisien. Teknologi blockchain, terutama melalui kontrak pintar, menyediakan metode terdesentralisasi dan tahan gangguan untuk mengelola data medis. Namun, menyimpan data skala besar langsung di blockchain menyebabkan biaya transaksi yang jauh lebih tinggi (biaya gas), membuat prosesnya tidak efisien secara ekonomi dan secara teknis tidak praktis. Untuk mengatasi keterbatasan ini, penelitian ini meningkatkan integrasi kontrak pintar blockchain dengan InterPlanetary File System (IPFS) dengan mengembangkan kerangka kerja manajemen data yang aman, andal, dan hemat biaya. Arsitektur hibrida yang diusulkan menggabungkan kontrak pintar on-chain untuk kontrol akses dan IPFS off-chain untuk penyimpanan data, secara efektif mengurangi biaya gas sekaligus menjaga keamanan dan transparansi. Hasil eksperimen menunjukkan bahwa pendekatan hibrida memastikan penanganan data medis yang tidak dapat diubah, dapat diverifikasi dan efisien. Evaluasi terhadap prototipe sistem yang ditawarkan mengkonfirmasi integritas data dan pengaksesan data dengan cepat. Analisis kinerja menunjukkan sistem berhasil mengurangi tantangan biaya, meningkatkan skalabilitas dan keamanan dalam manajemen data medis secara terdesentralisasi. Secure and reliable medical data management is a major challenge in digital healthcare systems, where the exponential growth of electronic medical records increases the demand for scalable and efficient storage solutions. Blockchain technology, particularly through smart contracts, provides a decentralized and tamper-resistant method for managing medical data. However, storing large-scale data directly on the blockchain leads to significantly higher transaction costs (gas fees), making the process economically inefficient and technically impractical. To address this limitation, this study investigates the integration of blockchain smart contracts with the InterPlanetary File System (IPFS) to develop a secure, reliable, and cost-effective data management framework. The proposed hybrid architecture combines on-chain smart contracts for access control and off-chain IPFS for data storage, effectively reducing gas costs while maintaining security and transparency. Experimental results show that the hybrid approach ensures immutable, verifiable, and efficient medical data handling. Field evaluations confirm data integrity and rapid retrieval, while performance analysis demonstrates that the model successfully mitigates cost, scalability, and security challenges in decentralized healthcare data management.
Strategi Retensi Pelanggan Berbasis Historis: Optimalisasi Model Prediksi Churn Menggunakan Machine Learning Abdul Wahid; Agung Muliawan
SemanTIK : Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi
Publisher : Informatics Engineering Department of Halu Oleo University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55679/semantik.v11i2.237

Abstract

Retensi pelanggan merupakan aspek strategis yang sangat penting dalam menjaga keberlanjutan bisnis, terutama di era kompetisi yang semakin ketat. Penelitian ini berfokus pada upaya optimalisasi model prediksi customer churn berbasis riwayat historis pelanggan dengan memanfaatkan pendekatan machine learning. Dua algoritma utama yang digunakan adalah Support Vector Machine (SVM) dan Jaringan Saraf Tiruan (Artificial Neural Network/ANN) sebagai representasi dari metode ANN. Untuk meningkatkan performa prediksi, diterapkan pula teknik ensemble classifier berupa bagging dan boosting. Guna mengatasi kompleksitas data dan mengurangi risiko overfitting, digunakan teknik dimensionality reduction melalui Principal Component Analysis (PCA). Dataset yang digunakan mencakup berbagai variabel penting seperti data demografis, perilaku pembelian, serta interaksi pelanggan dengan perusahaan. Hasil penelitian menunjukkan bahwa penerapan PCA mampu meningkatkan akurasi model, di mana ANN mencapai 92,37% dan SVM 85,13%. Penerapan metode boosting meningkatkan performa menjadi 93,34% untuk ANN dan 92,73% untuk SVM, sedangkan hasil terbaik diperoleh melalui bagging dengan akurasi 94,38% dan 94,15%. Temuan ini membuktikan bahwa kombinasi antara reduksi dimensi dan ensemble classifier dapat secara signifikan meningkatkan ketepatan prediksi customer churn, sehingga mendukung pengambilan keputusan strategis dan penyusunan strategi retensi pelanggan yang lebih proaktif, terukur, dan efektif. Customer retention is a very important strategic aspect in maintaining business sustainability, especially in an era of increasingly fierce competition. This study focuses on optimizing the customer churn prediction model based on customer historical data by utilizing a machine learning approach. The two main algorithms used are Support Vector Machine (SVM) and Artificial Neural Network (ANN) as representations of ANN methods. To improve prediction performance, ensemble classifier techniques such as bagging and boosting were also applied. To overcome data complexity and reduce the risk of overfitting, dimensionality reduction techniques were used through Principal Component Analysis (PCA). The dataset used included various important variables such as demographic data, purchasing behavior, and customer interactions with the company. The results show that the application of PCA improves model accuracy, with ANN reaching 92.37% and SVM 85.13%. The application of the boosting method improves performance to 93.34% for ANN and 92.73% for SVM, while the best results are obtained through bagging with an accuracy of 94.38% and 94.15%. These findings prove that the combination of dimension reduction and ensemble classifiers can significantly improve the accuracy of churn prediction, thereby supporting strategic decision-making and the development of more proactive, measurable, and effective customer retention strategies.

Page 5 of 7 | Total Record : 64

 3   4   5   6   7 

Filter by Year

2024 2026


Reset
Filter By Issues
All Issue Vol. 12 No. 1 (2026): Volume 12 Number 1 (January-june 2026) Vol. 11 No. 1 (2025): Vol. 11 No. 1 (2025): SemanTIK Teknik Informasi Vol. 11 No. 2 (2025): SemanTIK : Teknik Informasi Vol. 10 No. 2 (2024): SemanTIK : Teknik Informasi