The Department of Statistics, Universitas Padjadjaran, has once again contributed to innovative research published in the international journal Communications in Mathematical Biology and Neuroscience. The paper, titled “Emotion classification in social media posts related to telecommunication services using bidirectional LSTM”, was written by Najwa et. al (2025)
This study highlights the rapid growth of social media usage in Indonesia, which reached 139 million users as of January 2024. Social media platforms have become a primary channel for expressing public opinion, including customer experiences with telecommunication services. The research applies Bidirectional Long Short-Term Memory (BiLSTM) with FastText word embeddings to classify emotions in posts related to major Indonesian telecom providers, such as Telkomsel, Indosat, XL, and Axis.
By implementing the Borderline-SMOTE technique to address class imbalance, the model successfully categorized six basic emotions: happiness, sadness, fear, anger, disgust, and surprise. The proposed model achieved impressive results, with 93.51% accuracy and an F1-score of 93.48% on unseen data.
Beyond its academic contribution in the fields of data science and text analytics, this research provides practical insights for the telecommunication industry to better understand customer sentiment, improve service quality, and strengthen engagement with the community.
The study also aligns with the global Sustainable Development Goals (SDGs), particularly:
- SDG 9 (Industry, Innovation, and Infrastructure): advancing the use of smart technologies to improve public services.
- SDG 16 (Peace, Justice, and Strong Institutions): supporting transparency and citizen-centered services.
Through this publication, the Department of Statistics at Universitas Padjadjaran reaffirms its commitment to connecting academic research with real-world needs, while actively contributing to Indonesia’s progress toward achieving the SDGs.
Source: https://www.scopus.com/record/display.url?eid=2-s2.0-85217203748&origin=resultslist
05/Stat/2025
