Claim Missing Document
Check
Articles

Analisis Sentimen Komentar YouTube Terhadap Korporasi MBG: Studi Perbandingan Metode BERT dan Support Vector Machine (SVM) Najma Alfisyahrina; Riza Ibnu Adam; Aries Suharso
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.6822

Abstract

Program Makan Bergizi Gratis (MBG) merupakan program pemerintah yang memperoleh berbagai tanggapan dari masyarakat melalui media sosial, khususnya YouTube. Komentar yang dihasilkan pengguna dapat dimanfaatkan untuk mengidentifikasi opini publik terhadap implementasi program tersebut. Penelitian ini bertujuan menganalisis sentimen publik terhadap program MBG serta membandingkan performa metode Bidirectional Encoder Representations from Transformers (BERT) dan Support Vector Machine (SVM) dalam klasifikasi sentimen. Data penelitian diperoleh dari komentar YouTube yang berkaitan dengan program MBG dan menghasilkan 4.109 komentar valid setelah melalui tahap preprocessing. Model BERT diimplementasikan menggunakan BERT-Base-Uncased melalui proses fine-tuning, sedangkan SVM menggunakan pembobotan TF-IDF sebagai representasi fitur. Evaluasi model dilakukan menggunakan metrik accuracy, precision, recall, dan F1-score. Hasil penelitian menunjukkan bahwa metode BERT menghasilkan performa klasifikasi yang lebih baik dibandingkan SVM dalam mengidentifikasi sentimen positif, netral, dan negatif. Analisis temporal juga menunjukkan adanya perubahan distribusi sentimen publik sebelum dan sesudah terjadinya insiden yang berkaitan dengan program MBG. Temuan penelitian menunjukkan bahwa model berbasis Transformer lebih efektif dalam memahami konteks bahasa pada data media sosial dibandingkan metode machine learning tradisional.
Implementasi Metode Simple Additive Weighting (SAW) pada Sistem Informasi Magang Mahasiswa Berbasis Web untuk Penilaian Akhir Magang Annisya Dwi Chaerani; Aries Suharso; Asep Jamaludin
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7189

Abstract

The implementation of the student internship program at the Faculty of Computer Science, Universitas Singaperbangsa Karawang (Fasilkom UNSIKA) is currently hindered by a semi-conventional administrative system, decentralized academic document archiving, and vulnerability to human errors during the conversion score calculation process. This study aims to design and develop a web-based Student Internship Information System that specifically integrates a Decision Support System (DSS) function to digitalize the bureaucratic cycle while automating the internship final evaluation in an objective, transparent, and accurate manner. The software engineering applies the Waterfall development model and is constructed using a modern tech stack consisting of Next.js, React.js, Node.js, Sequelize ORM, and a MySQL database. To produce fair evaluation decisions, the platform implements the Simple Additive Weighting (SAW) algorithm within the backend system as a computational engine to execute benefit normalization and linear weighted summation of multi-source score components, aligned with the faculty's internship conversion Course Learning Outcomes (CPMK) criteria matrix. The result of this thesis research is a centralized digital platform that accommodates the administrative data needs of students, supervisors, examiners, and administrators. Through the implementation of the automated SAW algorithm, the system is proven reliable in streamlining the conventional grade recapitulation bureaucracy, eliminating human error risks, and presenting transparent score sheets along with precise evaluation ranking charts to enhance the faculty's academic management efficiency.
Implementasi Arsitektur OLAP pada Database OLTP dengan Log-Based CDC: Studi Kasus SIMKOPDES Yonatan Yusak Lestari; Oman Komarudin; Aries Suharso
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7199

Abstract

SIMKOPDES is an information system that supports the management and presentation of Koperasi Desa/Kelurahan Merah Putih information. Directly executing reporting workloads on the Online Transaction Processing (OLTP) database may create resource contention because analytical queries involve multiple joins, aggregations, and large-scale data processing. This study designs and implements an Online Analytical Processing (OLAP) architecture on the SIMKOPDES OLTP database using a log-based Change Data Capture (CDC) mechanism. The study adopts the Database System Development Life Cycle (DSDLC), while Kimball's Four-Step Dimensional Design is applied during database design. The OLAP schema consists of four dimension tables and three fact tables. Data changes are synchronized using PostgreSQL write-ahead logging, a replication slot, the wal2json plugin, and Python middleware. Performance evaluation using EXPLAIN (ANALYZE, BUFFERS) across five analytical scenarios yielded average efficiencies of 85.23% in execution time, 53.59% in I/O buffer usage, and 80.09% in query cost relative to OLTP. Across all five scenarios, the OLAP row counts and aggregate values matched the OLTP results. In addition, all seven tested data-change scenarios were successfully captured and applied to the affected OLAP tables. These results indicate that an OLAP architecture with log-based CDC can improve analytical query efficiency while maintaining consistency with the OLTP source data.
Prinsip Nielsen pada Redesign User Experience SIMKOPDES Bintang Danuarta; Oman Komarudin; Aries Suharso
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7206

Abstract

Sistem Informasi Koperasi Desa Merah Putih (SIMKOPDES) is a digital platform to modernize cooperative governance in rural areas. However its implementation faces challenges due to the digital literacy gap among administrators and an existing interface design that triggers cognitive overload. This study aims to redesign the SIMKOPDES interface using the Design Thinking method. The problem identification stage was conducted comprehensively using Heuristic Evaluation (HE) based on Nielsen 10 Principles by experts and Cognitive Walkthrough (CW) by end users. The redesign solutions focused on simplifying navigation flows, adjusting terminology to match the mental models of rural users, and providing clear visual feedback. Final testing of the new prototype was conducted using the UEQ instrument with 20 rural cooperative administrators as respondents. The measurement results showed a highly significant improvement with all scales achieving positive values. Overall, the evaluation results demonstrate a substantial improvement in user experience, with an average UEQ score increase of 3.04 points from the existing system to the redesigned prototype. The most significant improvement was recorded in the Perspicuity dimension, which increased from -0.89 to 2.05, indicating that the prototype is considerably easier to understand and learn for users. In the benchmark comparison five scales were categorized as Excellent and the Novelty scale as Good. This proves that user experience engineering based on heuristic principles successfully created an adaptive, friendly, and inclusive SIMKOPDES interface for cooperative administrators with limited digital literacy.
Implementasi Fisher-Yates Shuffle pada Fitur Kuis dalam Media Pembelajaran EduBidan Berbasis Web (Studi Kasus: Mahasiswa Kebidanan Unsika) Muhamad Ikhsan Rizqi Yanuar; Aries Suharso; Asep Jamaludin
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7207

Abstract

Midwifery education requires structured, flexible, and user-friendly digital media to support conceptual, procedural, and visual learning as well as assessments with varied question orders. This study develops EduBidan as a web-based Digital Midwifery LMS, analyzes Fisher-Yates Shuffle in evaluation quizzes, and evaluates functionality and usability. The system was developed using the Prototype method through five stages and three evaluation iterations. EduBidan supports Student, Lecturer, and Administrator roles with features for materials, quizzes, progress tracking, and grade summaries. Fisher-Yates Shuffle randomizes questions and answer options using a seed to maintain consistent arrangements for the same student and quiz. Testing included Black Box Testing, algorithm testing, and the System Usability Scale (SUS). All Black Box scenarios produced the expected results. Three algorithm-testing scenarios showed reordered items without missing or duplicated data, consistency after page reloads, and accurate scoring. SUS testing with 34 respondents produced a score of 84.41, grade B, an adjective rating of Excellent, and an acceptability range of Acceptable. The results indicate that EduBidan fulfills functional requirements, has good usability, and is acceptable to users.
PREDIKSI PRODUKSI PADI KABUPATEN KARAWANG BERDASARKAN LUAS PANEN PADI MENGGUNAKAN CITRA SATELIT Rizky Afriza Rosal Fino; Aries Suharso; Chaerur Rozikin
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7222

Abstract

Kabupaten Karawang merupakan salah satu lumbung padi utama di Provinsi Jawa Barat, tetapialih fungsi lahan dan dinamika iklim membuat pemantauan produksi padi yang akurat dan tepatwaktu menjadi kebutuhan yang semakin mendesak. Penelitian ini bertujuan membangun danmengevaluasi model prediksi produktivitas padi (ton/ha) berbasis citra satelit multisensor danmachine learning pada 30 kecamatan di Kabupaten Karawang selama periode 2019 sampai 2025,dengan mengadopsi kerangka CRISP-DM. Fitur masukan diekstraksi dari citra Sentinel-2 (NDVIdan LSWI), Sentinel-1 SAR, serta data iklim pendukung melalui Google Earth Engine. Limaalgoritma diuji secara komparatif menggunakan skema validasi Leave-One-Year-Out (LOYO),yaitu Random Forest, Ordinary Least Squares (OLS), Ridge Regression, XGBoost, danLightGBM. Random Forest menjadi model terbaik pada level kecamatan dengan MAPE 7,207%,RMSE 0,580 ton/ha, dan Akurasi Kabupaten 94,154%, dengan R² sebesar 0,961 pada skalaproduksi absolut tingkat kecamatan. Pada level kabupaten, OLS tampil lebih unggul denganMAPE 5,279%. Saat diterapkan untuk estimasi produksi tahun 2025, model menghasilkan prediksi1.475.946,3 ton dibandingkan data aktual 1.439.910,2 ton, setara Akurasi Kabupaten 97,50%.Temuan ini menegaskan bahwa model berbasis citra satelit dapat difungsikan sebagai instrumenpendukung estimasi produktivitas dan produksi padi di Kabupaten Karawang