Analisis Sentimen Kecelakaan Lalu Lintas Di Tweet Twitter Menggunakan Metode Naïve Bayes

Aldy Sahputra Saragih (Universitas Mandiri Bina Prestasi)
Evantrana Yordan Bangun (Universitas Mandiri Bina Prestasi)
Fastabikha Akbar Fahlevi (Universitas Mandiri Bina Prestasi)
Ariel Kurniawan (Universitas Mandiri Bina Prestasi)
Kristian Sigalingging (Universitas Mandiri Bina Prestasi)
Fasta Krel Four Wati Br Haloho (Universitas Mandiri Bina Prestasi)
Salman Putra Jaya Hulu (Universitas Mandiri Bina Prestasi)

Scripta Technica: Journal of Engineering and Applied Technology
Vol 2 No 1 (2026): : June: Scripta Technica: Journal of Engineering and Applied Technology

This study develops a robust computational framework using a structured Naïve Bayes architecture to mine, classify, and analyze public discourse regarding traffic accidents on Twitter (now X). Utilizing a massive corpus of 157,629 digital documents, the probabilistic model successfully extracts macro-level sentiment distributions with a high global accuracy of 82.60%. The empirical findings reveal an overwhelming dominance of external factor attributions, accumulating 128,613 tweets, which underscores a critical public sensitivity toward infrastructure malfunctions and suboptimal road conditions in urban environments. By transforming unorganized microblogging texts into structured analytical matrices, this research establishes an economical, real-time social sensor that effectively bypasses the logistical delays of conventional manual reporting systems. Ultimately, this study contributes a novel theoretical repositioning from reactive safety evaluations to data-driven preventive strategies, providing government authorities with a rigorous, scalable decision-making tool to optimize sustainable transportation infrastructure and mitigate urban traffic fatalities.