Sentiment Analysis for Public Services Using Bidirectional Long Short-Term Memory Method

Prasetyo, Vincentius Riandaru and Naufal, Mohammad Farid and Soeratman, Nelson (2025) Sentiment Analysis for Public Services Using Bidirectional Long Short-Term Memory Method. In: 2025 International Seminar on Intelligent Technology and Its Applications (ISITIA), 23-25 Juli 2025, Surabaya.

PDF Vincent_ISITIA 2025.pdf - Published Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Sentiment analysis of public services based on social media has great potential to evaluate the quality of government services automatically. However,the main challenge is the complexity of opinion data, which is making it difficult to classify accurately by classical methods. This study aims to overcome this problem by applying a deep learning method based on Bidirectional Long Short-Term Memory (Bi-LSTM) to analyse the sentiment of public opinion in Surabaya about public services taken from social media platforms X(Twitter), Instagram, and GoogleReviews. The test results showed that the proposed Bi-LSTM model achieved an accuracy of94.6%, and precision,recall,and F1-score values o f95.6%,91.7%, and 93.6%,respectively.These findings prove that the Bi-LSTM model can capture the context of two-way sentences more effectively than classical algorithms and provides novelty in utilizing multplatform datasets in the public service domain for Surabaya City. This research offers practical implications in the form of automated methods that can assist the government in data-based decision making to improve the quality of public services.

Item Type:	Conference or Workshop Item (Paper)
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software
Divisions:	Faculty of Engineering > Department of Informatic
Depositing User:	VINCENTIUS RIANDARU PRASETYO
Date Deposited:	02 Oct 2025 05:16
Last Modified:	02 Oct 2025 05:16
URI:	http://repository.ubaya.ac.id/id/eprint/49635

Actions (login required)

View Item