Garuda - Garba Rujukan Digital

Prediksi Indeks Prestasi Komulatif Mahasiswa berdasarkan Nem dengan Menggunakan Algoritma Neural Network Berbasis Particle Swarm Optimization : Prediction of Student Comulative Achievement Index Based on NEM Using Particle Swarm Optimization Based Neural Network Algorithm Muhamad Ziaul Haq; Nursalim
Jurnal Kolaboratif Sains Vol. 6 No. 2: FEBRUARI 2023 - Jurnal Kolaboratif Sains (JKS)
Publisher : Universitas Muhammadiyah Palu

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56338/jks.v6i2.3303

Proses Penerimaan Mahasiswa Baru (PMB) pada berbagai perguruan tinggi di Indonesia baik itu perguruan tinggi negeri ataupun swasta melakukan seleksi terhadap calon mahasiswanya dengan melihat pada Nilai Ebtanas Murni (NEM). Guna menganalisis hubungan antara nilai NEM calon mahasiswa dengan prestasi akademik yang dicapai di STMIK Adhi Guna (dalam hal ini digunakan indeks prestasi kumulatif (IPK) dengan analisis korelasi dan regresi linier ganda. Variabel penelitian yang digunakan adalah hasil IPK kelulusan mahasiswa sebagai variabel dependen (terikat), dan nilai mata pelajaran Bahasa Indonesia, Bahasa Inggris, dan Matematika sebagai variabel independent (bebas). Penelitian ini menggunakan 2 algoritma yang berbeda yaitu Neural Network (NN) dan Particle Swarm Optimization (PSO) untuk membandingkan hasil dan mencari tingkat akurasi terbaik diantara kedua algoritma tersebut. Dari penelitian yang telah dilakukan dapat disimpulkan bahwa Neural Network berbasis Particle Swarm Optimization adalah algoritma yang paling baik dibandingkan dengan Neural Network untuk mengukur tingkat korelasi antara NEM dan Indeks Prestasi Kumulatif (IPK) mahasiswa. Neural Network pada penelitian ini menghasilkan akurasi terkecil 0,214 dengan Time = 4 s. Neural Network berbasis PSO menghasilkan akurasi terkecil 0,132.

The Implementation of Simple Additive Weighting Method for Designing A Web-Based Waste Management Saving Transaction System Nursalim; Muhamad Ziaul Haq; Nalis Hendrawan; Roy Mubarak; I Putu Dody Suarnatha
Jurnal Sistim Informasi dan Teknologi 2023, Vol. 5, No. 2
Publisher : SEULANGA SYSTEM PUBLISHER

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37034/jsisfotek.v5i2.244

The Garbage Bank is an organization initiated for all students to support the management of inorganic waste into something of value so as to create additional income. The increasing number of customers has caused the treasurer to be overwhelmed in ranking customer rankings and to not be on target in determining the best customer with the same amount of waste. In addition, there is no system security in handling the transaction process, so unwanted access can occur. The purpose of this research is to develop a savings transaction system for waste management so that it becomes a green Campus. The decision-making method uses Simple Additive Weighting. The system development methodology used is Rapid Application Development (RAD). The tools used in system design are the Unified Modeling Language. The implementation of this system uses the PHP programming language with the Laravel and MySQL frameworks for database processing. The resulting system can simplify and speed up the process of recording and managing waste bank data.

The Application of Information Technology Architectural Design Using TOGAF Architecture Framework in Restaurant Service Systems Sri Wahyuningsih, Suluh; Ziaul Haq, Muhammad; Hamid, Helson; Hady, Sultan; Hendrawan, Nalis
Jurnal Informasi dan Teknologi 2023, Vol. 5, No. 4
Publisher : SEULANGA SYSTEM PUBLISHER

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.60083/jidt.v5i4.429

This research aims to see how TOGAF ADM is applied to modeling information technology architecture in restaurants. In this research, the author used the TOGAF Architecture Development Method (ADM). In TOGAF ADM, there is a definition of architecture and its understanding, which is in the preliminary phase (the preparatory phase). In this modeling, it starts from zero, so a detailed architectural process is needed. This is needed to simplify the subsequent architectural development process. Detailed architectural processes can be obtained using the framework. In TOGAF ADM, there are stages that have been arranged in such a way that the details of the architecture can be seen in them. The modeling that the author will compile also requires support for architectural evolution. This is needed because, initially, the restaurant did not have technological architecture. In Phase F Migration Planning, the framework provides support for technology architecture evolution. Based on the research steps, there are 8 structured stages plus a preliminary stage. However, in this research, the author will only discuss up to stage F, namely migration planning. From the research conducted by the author, a model of information technology architecture for restaurants was obtained, which was implemented using TOGAF ADM. The information technology architecture model includes service processes, payment processes, and monitoring processes.

Application of the K-Nearest Neighbor Algorithm Method to Analyze Netizen Responses and Reactions Toward the Relocation of Capital City at social media Ziaul Haq, Muhammad; Sri Wahyuningsih, Suluh; Nursalim; Nuryanto, Uli Wildan; Rachman, Andy
Jurnal Informasi dan Teknologi 2023, Vol. 5, No. 4
Publisher : SEULANGA SYSTEM PUBLISHER

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.60083/jidt.v5i4.458

Sentiment analysis is a form of natural language processing that uses word analysis to ascertain people's thoughts, feelings, and views on a certain topic. In this study, word processing refers to the procedure used to categorize written texts into positive and negative emotion categories. Using data crawling techniques, information on public comments on the relocation of Indonesia's capital was gathered from Twitter social media. Keywords related to the move included "new capital," "moving capital," and "moving capital with 10,000 comments." The author of this work classified test data and training data using a lexical approach using the K-Nearest Neighbor (K-NN) method. The purpose of this study is to evaluate the K-NN algorithm's accuracy, error rate, precision, f-measure, and recall. In order to identify the ideal parameters, tests were also conducted on calculating the k value in the K-Nearest Neighbor (K-NN) method. Testing the K-Nearest Neighbor (K-NN) method yielded the greatest accuracy level of 60% with a k value of 9, concluding with the initial data collection. The K-Nearest Neighbor (K-NN) technique was evaluated in the second data collection, and with a k value of 5, it had the best accuracy level of 70%. Future scholars might create texts in languages other than Indonesian and categorize those that include visuals in them. Next, add more dictionaries to the collection and extract features from bigrams, trigrams, quadgrams, and other combinations. You may then employ several algorithmic techniques in the accuracy calculation feature.

Deep Learning Based Augmented Reality for 3D Object Recognition Muhamad Ziaul Haq; Nursalim Nursalim
International Journal Software Engineering and Computer Science (IJSECS) Vol. 5 No. 3 (2025): DECEMBER 2025
Publisher : Lembaga Komunitas Informasi Teknologi Aceh (KITA)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35870/ijsecs.v5i3.5431

Augmented Reality (AR) technology is being widely adopted in various fields such as education, entertainment, and creativity. However, there are still some challenges to be overcome in recognizing and rendering three-dimensional (3D) objects accurately and in real-time. We implemented an AR system that utilizes deep learning techniques to recognize 3D objects with improved accuracy levels. Our approach involved training a Convolutional Neural Network (CNN) model using 3D object datasets captured from different viewpoints. The development included designing the network architecture, training the model, evaluating its accuracy, and integrating it into an AR platform based on Unity 3D and Vuforia SDK. The results indicated that the system could achieve recognition of the 3D objects with an average accuracy of 93.7%, precision of 92.4%, and recall of 91.8%, all while keeping response times below 0.8 seconds. Objects with complex geometries like cars and chairs had recognition rates above 94%, while those with similar textures had lower accuracy because of detailed surface complexities. It allows stable interactive visualization of objects in augmented reality even under different lighting conditions and camera angles. Combining deep learning with AR improves the quality of object recognition and provides a more realistic interactive experience. This paper discusses the advances made in AR technology toward better adaptability and efficiency, which can be applied to interactive education, industrial simulation, architecture, and medical fields.

Algoritma Naïve Bayes untuk Mengidentifikasi Hoaks di Media Sosial Muhamad Ziaul Haq; Cut Susan Octiva; Ayuliana Ayuliana; Uli Wildan Nuryanto; Dikky Suryadi
Jurnal Minfo Polgan Vol. 13 No. 1 (2024): Artikel Penelitian
Publisher : Politeknik Ganesha Medan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33395/jmp.v13i1.13937

Hoaks yang menyebar di dunia maya terus bertambah, oleh sebab itu penelitian ini akan membuat dan memeriksa fungsionalitas dari algoritma Naïve Bayes dalam mendeteksi serta mengenali hoaks pada jejaring sosial. Algoritma yang diambil adalah Naïve Bayes karena dapat memproses data teks yang large-scale dengan kompleksitas tersendiri dan pengimplementasiannya yang mudah. Dataset yang digunakan untuk penelitian ini terdiri dari postingan media sosial yang dikategorikan sebagai hoaks atau bukan hoaks. Pra-pemrosesan data mencakup tokenisasi, ekstraksi fitur, dan pembersihan teks menggunakan teknik TF-IDF (Term Frequency-Inverse Document Frequency). Selanjutnya, algoritma Naïve Bayes dilatih dan diuji menggunakan cross-validation untuk memastikan bahwa model itu akurat dan dapat digunakan di mana saja. Hasil penelitian menunjukkan bahwa algoritma Naïve Bayes dapat mengidentifikasi hoaks dengan tingkat kesalahan yang sangat rendah. Penelitian ini menunjukkan bahwa algoritma Naive Bayes efektif dalam mendeteksi konten hoaks di media sosial, seperti yang ditunjukkan oleh evaluasi model menggunakan metrik seperti ketepatan, recall, dan skor F1. Penelitian ini juga menemukan bahwa algoritma ini dapat diintegrasikan dalam sistem pemantauan media sosial untuk meningkatkan kualitas informasi yang beredar.

AUTOMATED ESSAY SCORING FOR STUDENT EXAMS USING DEEP NLP MODELS Andi Kaimuddin; Bryant Ritchie Trisnodjojo; Muhamad Ziaul Haq; Nursalim; Riezky Purnama Sari
JTH: Journal of Technology and Health Vol. 4 No. 1 (2026): July: JTH: Journal of Technology and Health
Publisher : CV. Fahr Publishing

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61677/jth.v4i1.858

The increasing use of essay-based examinations in higher education has created significant challenges in maintaining efficient, objective, and consistent assessment processes. Manual essay grading is time-consuming and susceptible to subjective judgment, particularly when evaluating large numbers of student responses. Therefore, this study aims to develop and evaluate an Automated Essay Scoring (AES) system based on Bidirectional Encoder Representations from Transformers (BERT) to improve the accuracy and consistency of student essay assessment. This research employed an experimental quantitative approach using 2,500 student essay responses, of which 2,340 valid responses were retained after preprocessing and data cleaning. The dataset was divided into training, validation, and testing subsets using a 70:15:15 ratio. The proposed model was fine-tuned using the AdamW optimizer with a learning rate of 2 × 10⁻⁵, a batch size of 16, and 8 training epochs. Model performance was evaluated using Quadratic Weighted Kappa (QWK), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The experimental results demonstrate that the proposed BERT model achieved a QWK score of 0.872, indicating strong agreement with human evaluators, while obtaining an MAE of 0.418 and an RMSE of 0.593, reflecting relatively low prediction errors. Comparative evaluation also showed that the proposed BERT model outperformed conventional baseline approaches in automated essay scoring, confirming the effectiveness of contextual language representations for understanding semantic information in student essays. These findings indicate that the proposed framework provides a reliable and efficient solution for automated essay assessment, offering practical benefits for improving scoring consistency, reducing lecturers' workload, and supporting the implementation of intelligent assessment systems in higher education.

Deep Learning Based Augmented Reality for 3D Object Recognition Muhamad Ziaul Haq; Nursalim Nursalim
International Journal Software Engineering and Computer Science (IJSECS) Vol. 5 No. 3 (2025): DECEMBER 2025
Publisher : Lembaga Komunitas Informasi Teknologi Aceh (KITA), Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35870/ijsecs.v5i3.5431

Augmented Reality (AR) technology is being widely adopted in various fields such as education, entertainment, and creativity. However, there are still some challenges to be overcome in recognizing and rendering three-dimensional (3D) objects accurately and in real-time. We implemented an AR system that utilizes deep learning techniques to recognize 3D objects with improved accuracy levels. Our approach involved training a Convolutional Neural Network (CNN) model using 3D object datasets captured from different viewpoints. The development included designing the network architecture, training the model, evaluating its accuracy, and integrating it into an AR platform based on Unity 3D and Vuforia SDK. The results indicated that the system could achieve recognition of the 3D objects with an average accuracy of 93.7%, precision of 92.4%, and recall of 91.8%, all while keeping response times below 0.8 seconds. Objects with complex geometries like cars and chairs had recognition rates above 94%, while those with similar textures had lower accuracy because of detailed surface complexities. It allows stable interactive visualization of objects in augmented reality even under different lighting conditions and camera angles. Combining deep learning with AR improves the quality of object recognition and provides a more realistic interactive experience. This paper discusses the advances made in AR technology toward better adaptability and efficiency, which can be applied to interactive education, industrial simulation, architecture, and medical fields.

User Satisfaction Classification of Tiktok Shop Skincare Products Using C4.5 and Random Forest for Recommendation Strategy Nursalim Nursalim; Muhamad Ziaul Haq; A. Nurul Hidayat; Budi Mulyono
Sharia Economic and Management Business Journal (SEMBJ) Vol. 7 No. 2 (2026): Sharia Economic and Management Business
Publisher : Yayasan Darussalam Bengkulu

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62159/sembj.v7i2.2234

Background: TikTok Shop has become an important social commerce platform for skincare purchases; however, product recommendations are not always perceived as relevant by users. A data-driven satisfaction classification model is therefore needed to support more targeted recommendation strategies. Method: This study used a quantitative approach involving 150 TikTok Shop users who had purchased skincare products. Data were collected through an online questionnaire containing 14 Likert-scale items and three recommendation-preference items. Instrument quality was evaluated using corrected item-total correlation and Cronbach Alpha. The C4.5 decision tree and Random Forest models were evaluated using stratified 10-fold cross-validation. Results: All 14 items were valid, with item-total correlations ranging from 0.619 to 0.881, and the overall Cronbach Alpha was 0.969. The satisfaction classes were balanced, consisting of 75 satisfied and 75 unsatisfied respondents. Information gain analysis identified product delivery as the most influential attribute, with a gain value of 0.4551. C4.5 achieved 85.33% accuracy, while Random Forest achieved 83.33% accuracy. Conclusion: C4.5 provided competitive performance and stronger interpretability than Random Forest for this dataset. The resulting classification rules can be used to prioritize delivery reliability, application usability, and product quality in skincare recommendation strategies.

Title

Found 9 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 9 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 9 Documents
Search