PUBLIC SENTIMENT ANALYSIS ON TOLL ROAD DEVELOPMENT IN INDONESIA

• Main Author: Subhakti, Aditya
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79746
• Institution: Institut Teknologi Bandung

Public infrastructure plays a crucial role in the economic development of a country. One of the key areas focused on in economic development is toll road infrastructure. The establishment of toll road infrastructure has direct and indirect impacts, both positive and negative, on the surrounding social, economic, and environmental aspects. These impacts trigger varied reactions, both supportive and opposing, from the community towards toll road development. In Indonesia, there are 21.2 million users of the Twitter/X social media platform. Twitter/X offers rapid information dissemination through micro-blogging, where users can create short messages of up to 280 characters (tweets). The societal responses on social media platforms, particularly regarding infrastructure, specifically toll roads, allow for sentiment analysis to identify valuable research gaps and challenges. This contributes to knowledge enhancement for sustainable toll road development. Sentiment analysis involves studying opinions in text data using Natural Language Processing (NLP), a field of Artificial Intelligence (AI) that applies Machine Learning (ML). Within sentiment analysis, text is classified into positive, neutral, and negative sentiments. Machine learning is employed using the Naïve Bayes classification method to classify sentiment from tweet text data. The social impacts include hindered community accessibility, improved mobility, land use changes in settlements and livelihoods, and increased job opportunities. Economically, toll roads stimulate inter-regional economic activities, foster business district development, trigger residential area growth, promote tourism development, and save transportation time. Environmentally, toll road development leads to reductions in agricultural land, decreased water infiltration areas, disrupted irrigation routes, noise pollution, air contamination, increased surface water flow, disturbances to nature reserves, vegetation alterations, and soil changes. A total of 899,296 tweets with the keyword "toll road" on Twitter/X were collected from 2018 to 2022. Text cleaning was performed to eliminate URLs, mentions, hashtags (#), emojis, punctuation, numbers, stop words, and duplicate sentences or words using RStudio. This cleaning resulted in 494,616 refined tweets. The creation of the machine learning model began with manual sentiment labeling by five participants for 25,808 tweets. Using RStudio, a machine learning model for sentiment classification was developed with 97.3% accuracy aligned with human sentiment understanding. This model was then used to classify sentiments in the entire 494,616 tweets. The sentiment results were visualized in time-series graphs and categorized based on toll road development impact points as per literature studies. Analysis was conducted concerning significant events or periods and impact point analyses. Bar charts were employed, along with word clouds to highlight topics reacted upon by the public during specific periods or impacts. A fishbone diagram was created to analyze the cause-and-effect relationships of positive and negative sentiments in society. The analysis concluded that public sentiment towards toll roads fluctuates from positive to negative over time. Positive sentiment factors included toll roads triggering economic growth, enhancing business districts, fostering tourism, increasing inter-regional mobility, improving job opportunities, integrating with other public infrastructures, and satisfaction with services/operations. Negative sentiment factors stemmed from hindered community accessibility, land acquisition issues impacting livelihoods, lack of compensation and education, social issues related to disparities, cultural changes, dissatisfaction with government policies, environmental concerns leading to reduced green spaces, natural disasters, and dissatisfaction with services/operations. The most reacted topics by the public were related to economic aspects, signifying high awareness of toll road infrastructure's importance for economic growth. Conversely, there was less reaction concerning environmental aspects, indicating lower public awareness of toll road impacts on the environment, particularly regarding air pollution, noise, and land contour changes. Evaluation and proposals findings from this study involve planning considerations, emphasizing pedestrian access and community mobility to minimize disturbances caused by toll road construction, implementing sustainable toll road development to avoid environmental issues, and paying more attention to irrigation routes and slopes to reduce floods and landslides that can hinder toll road operations. From a social perspective, there's a need for better communication to educate about long-term toll road benefits, effective two-way communication between the government and affected communities to minimize social, economic, and environmental grievances. The sentiment analysis process and results provide an overview of public reactions to toll road development. Visual representations like word clouds and fishbone diagrams make it easier for planners or decisionmakers to draw conclusions about public concerns as toll road users, facilitating appropriate responses. Sentiment analysis aids in swift and informed decisionmaking for toll road planners and maintainers.