A Hybrid CNN-LSTM Approach for Sentiment Analysis of Poverty Issues in East Java

Abstract

Poverty remains a significant social issue in East Java, Indonesia, often discussed actively on social media platforms such as Twitter. This study aims to analyze public sentiment related to poverty by implementing a hybrid deep learning model that combines Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM). A dataset of tweets was collected using keyword-based crawling techniques, followed by a series of preprocessing steps including case folding, tokenization, filtering, and labeling. The hybrid CNN-LSTM model was then trained to classify sentiments into positive and negative categories. Evaluation results show that the proposed model achieved a high accuracy of 91%, with equally strong performance in precision, recall, and F1-score. The findings also revealed that "unemployment" was the most dominant topic associated with poverty. This research demonstrates that the hybrid CNN-LSTM architecture is effective for sentiment analysis on short, informal text, and offers valuable insights into public opinion for policymakers.