Article Per Year (5 Year)
p-Index From 2021 - 2026
0.61
P-Index
This Author published in this journals
All Journal JOURNAL OF INFORMATION SYSTEM RESEARCH (JOSH) Bulletin of Computer Science Research
Arif Siswandi
Universitas Pelita Bangsa, Bekasi
Computer Science & IT Decision Sciences, Operations Research & Management

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : bulletin of computer science research

 1 
Identifikasi Indikasi Risiko Depresi pada Unggahan Media Sosial X Menggunakan Natural Language Processing dan Algoritma Random Forest Arif Siswandi; Arif Susilo
Bulletin of Computer Science Research Vol. 6 No. 2 (2026): February 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v6i2.1026

Abstract

Depression among university students has become an important mental health concern due to its potential impact on quality of life and academic performance. Social media platform X, as a text-based communication medium, provides a space for spontaneous expression that may reflect users’ emotional states. This study aims to analyze linguistic patterns associated with indicative depressive expressions in social media posts using a Natural Language Processing (NLP) approach and the Random Forest algorithm. Data were collected through web scraping between January and November 2024 using keywords conceptually derived from the Patient Health Questionnaire-9 (PHQ-9) indicators and adapted to linguistic expressions commonly used in social media communication. From an initial collection of 36,081 posts, several filtering stages were conducted, including duplicate removal, language filtering, and elimination of irrelevant content, resulting in a final dataset of 1,070 posts used in this study. The high filtering rate indicates that many scraped posts did not directly represent relevant emotional expressions. The dataset was manually labeled into three indicative categories of depressive expressions: mild, moderate, and severe. The analytical process included text preprocessing, TF-IDF feature extraction, and classification modeling using the Random Forest algorithm. The evaluation results show an accuracy of 97%. However, this value should be interpreted cautiously because model performance may be influenced by dataset characteristics and the manual labeling process. Therefore, the proposed model should be regarded as an exploratory approach for identifying linguistic patterns associated with emotional expressions in social media text rather than a clinical diagnostic tool for depression.
Co-Authors Arif Susilo Rizki Muhammad Mukti Sufajar Butsianto