Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Jurnal Teknologi Sistem Informasi Journal of Novel Engineering Science and Technology
Muhammad Fermi Pasha
Universitas Mikroskil
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Environmental Science Mechanical Engineering

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Optimization of hybrid-based Collaborative Filtering using Matrix Factorization, Feedforward Neural Network, and XGBoost Filimantaptius Gulo; Ronsen Purba; Muhammad Fermi Pasha
Journal of Novel Engineering Science and Technology Vol. 5 No. 02 (2026): Journal of Novel Engineering Science and Technology
Publisher : The Indonesian Institute of Science and Technology Research

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56741/jnest.v5i02.1356

Abstract

Collaborative filtering recommendation systems are widely used in digital applications; however, they still face challenges such as cold-start and first-rater problems, as well as limited accuracy due to their inability to capture complex user–item relationships. This study proposes a hybrid recommendation model that integrates Matrix Factorization, MLP-based Feedforward Neural Network (MLP) and Extreme Gradient Boosting (XGBoost). Experiments were conducted on two real-world datasets, namely MovieLens (movies) and PT XYZ (hotels), to validate the effectiveness of the proposed approach. The results indicate that the hybrid model consistently outperforms baseline methods such as SGD-based Matrix factorization, Matrix factorization +MLP, and user/item-based Collaborative filtering. Specifically, the integration of nonlinear learning through MLP and feature enhancement via XGBoost significantly improves prediction accuracy while mitigating cold-start and first-rater issues. These findings suggest that hybrid machine learning–based approaches can advance the development of more adaptive, accurate, and personalized recommendation systems.
Deteksi Ujaran Kebencian Dalam Domain Politik Pada Media Sosial Dengan Algoritma Long Short-Term Memory (LSTM) Alfiandri Putra Perdana; Ronsen Purba; Muhammad Fermi Pasha
Jurnal Teknologi Sistem Informasi Vol 7 No 2 (2026): Jurnal Teknologi Sistem Informasi (on Process)
Publisher : Program Studi Sistem Informasi, Universitas Multi Data Palembang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35957/jtsi.v7i2.17389

Abstract

Maraknya ujaran kebencian dalam domain politik di media sosial dapat memicu tindakan kekerasan, prasangka negatif bahkan perpecahan. Dengan metode pendeteksian ujaran kebencian dapat membantu menganalisis ujaran antara ujaran kebencian dan bukan ujaran kebencian. Namun, akurasi deteksi ujaran kebencian dibatasi oleh kualitas sumber dataset, penggunaan bahasa informal, definisi yang berbeda tentang apa yang dimaksud dengan ujaran kebencian. Penelitian ini memberikan solusi atas permasalahan ini menggunakan metode Long Short-Term Memory dan teknik pembobotan kata TF-IDF. Dataset yang digunakan akan melalui beberapa tahapan pre-processing, seperti Data Cleaning, Case Folding, Tokenizing, Filtering, dan Stemming. Setelah itu dataset akan dilakukan pembobotan dengan teknik TF-IDF. Kemudian seluruh data dilatih dan divalidasi dengan menggunakan model LSTM. Penelitian ini menghasilkan 12 model yang dilatih menggunakan dataset Alfina, Ibrohim, dataset hasil scrapping penulis, serta gabungan dataset tersebut yang kemudian dapat digunakan untuk menganalisis ujaran kebencian. Hasil pengujian menunjukkan proses deteksi ujaran kebencian lebih akurat, peningkatan akurasi dari 74,47% menjadi 96,58% dibandingkan dengan pengujian tanpa menerapkan metode usulan.
Co-Authors Alfiandri Putra Perdana Filimantaptius Gulo Ronsen Purba Ronsen Purba