Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Progresif: Jurnal Ilmiah Komputer Tematik : Jurnal Teknologi Informasi Komunikasi
Mohammad Zoqi Sarwani
Universitas Merdeka Pasuruan
Computer Science & IT Control & Systems Engineering

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

Found 2 Documents
Search

 1 
Text Analysis of Workplace Accident Chronology for Hazard Identification Using BERT and Random Forest Algorithms Mohammad Zoqi Sarwani; Anang Aris Widodo
TEMATIK Vol. 13 No. 1 (2026): Tematik : Jurnal Teknologi Informasi Komunikasi (e-Journal) - Juni 2026
Publisher : LPPM POLITEKNIK LP3I BANDUNG

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38204/tematik.v13i1.3004

Abstract

Workplace accidents in industrial environments continue to cause significant human, economic, and operational losses, making proactive hazard identification a critical priority for occupational health and safety (OHS) management. This study develops a workplace hazard-identification pipeline that automatically analyzes accident chronology texts using BERT (Bidirectional Encoder Representations from Transformers) as a contextual feature extractor and Random Forest as an ensemble classifier. The publicly available IHMStefanini Industrial Safety and Health dataset was used as the source corpus; after data-quality screening, stratified train-validation-test splitting (80:10:10), and class-imbalance handling using SMOTE applied only to BERT embedding vectors in the training partition, the final modeling matrix covered 14 Critical Risk hazard categories. Text preprocessing included controlled normalization, tokenization with bert-base-uncased, padding and truncation to 128 tokens, and contextual embedding extraction into 768-dimensional feature vectors. Experimental results on the held-out test set show that the proposed BERT-Random Forest model achieved an accuracy of 94.7%, precision of 93.8%, recall of 94.2%, and F1-score of 94.0%, outperforming TF-IDF with SVM, Word2Vec with LSTM, BERT with SVM, and standalone BERT fine-tuning baselines. Statistical comparison using the McNemar-Bowker paired error test confirmed that the performance difference between the proposed model and the strongest baseline was significant (p < 0.01). The main contribution of this study is not the generic superiority of a hybrid BERT-Random Forest architecture, but its practical adaptation for multi-class workplace hazard identification from industrial accident narratives with lower computational cost than full transformer fine-tuning. The proposed method can support automatic incident triage, hazard monitoring, prioritization of safety investigations, and decision.
KLASIFIKASI TEKS DEPRESI PADA MEDIA SOSIAL MENGGUNAKAN CONVOLUTION NEURAL NETWORK (CNN) DENGAN FASTTEXT EMBEDDING Azki Hamdi; Mohammad Zoqi Sarwani; Anang Aris Widodo
Progresif: Jurnal Ilmiah Komputer Vol. 22 No. 3 (2026): Juli
Publisher : STMIK Banjarbaru

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35889/progresif.v22i3.3899

Abstract

Depression is a persistent mood disorder that significantly impacts an individual's functioning, and its prevalence is becoming increasingly concerning among adolescents (the "Strawberry Generation"). Expressions of this psychological distress have largely migrated to social media in the form of unstructured text posts laden with informal terms, abbreviations, and typos. These linguistic characteristics pose "Out-of-Vocabulary" (OOV) challenges for traditional classification models. This study proposes integrating FastText word embeddings—which utilize subword information—to enhance feature representation, combined with a Convolutional Neural Network (CNN) architecture as the primary classifier. Using the "Student Depression Text" secondary dataset from Kaggle (comprising 7,489 entries), class imbalance was addressed exclusively within the training data using the Random Oversampling technique. Experimental results demonstrate that the proposed model achieved an overall accuracy of 95%, with Precision, Recall, and F1-Score values ​​for the depression class of 90%, 80%, and 84%, respectively. The integration of CNN and pre-trained FastText proved to be a reliable and efficient solution for managing the complexities of digital language in the context of early mental health detection. Keywords: Depression; Social Media; Natural Language Processing; Convolutional Neural Network; FastText Embedding. Abstrak Depresi merupakan gangguan suasana hati persisten yang secara signifikan memengaruhi fungsi fungsional individu dan prevalensinya kian mengkhawatirkan pada kelompok remaja (Strawberry Generation). Ekspresi tekanan psikologis ini kini banyak bermigrasi ke media sosial dalam bentuk unggahan teks tidak terstruktur yang sarat akan istilah non-formal, singkatan, dan salah ketik (typo). Karakteristik bahasa tersebut memicu kendala Out-of-Vocabulary (OOV) pada model klasifikasi tradisional. Penelitian ini mengusulkan integrasi word embedding FastText berbasis informasi tingkat sub-kata (subword information) untuk memperkuat representasi fitur, yang dikombinasikan dengan arsitektur Convolutional Neural Network (CNN) sebagai pengklasifikasi utama. Menggunakan dataset sekunder "Student Depression Text" sebanyak 7.489 entri dari Kaggle, ketidakseimbangan kelas diatasi secara eksklusif pada data latih menggunakan teknik Random Oversampling. Hasil eksperimen menunjukkan bahwa model yang diusulkan meraih akurasi keseluruhan sebesar 95%, dengan nilai Precision, Recall, dan F1-Score untuk kelas depresi masing-masing sebesar 90%, 80%, dan 84%. Integrasi CNN dan pre-trained FastText terbukti efektif menjadi solusi jalan tengah yang andal dan efisien dalam menangani kompleksitas bahasa digital untuk deteksi dini kesehatan mental.
Co-Authors Anang Aris Widodo Azki Hamdi