Subject: air pollution, pm10, western balkans, social media, tweets analysis, news media, sentiment analysis, topic modelling, cross correlation
Year: 2022
Type: Proceedings
Title: Investigating Public Awareness of Air Pollution in Western Balkans by analyzing Tweets and News Article Teasers
Author: Madjar, Angela
Author: Gjorshoska, Ivana
Author: Prodanova, Jana
Author: Dedinec, Aleksandra
Abstract: Air pollution is a serious threat to the health of people living in Western Balkans, with the biomass being one of the main pollutants. This study is based on the assumption that air pollution escalation in Western Balkan countries during winter will provoke a more intense activity on Twitter, while acknowledging that people’s opinions and feelings can often be influenced by content presented in news articles. The objective of this study is to investigate public awareness of air pollution in Macedonia, Serbia, Bosnia and Herzegovina and Montenegro. Natural Language Processing techniques such as Sentiment Analysis and Topic Modeling, as well as Statistical Analysis are employed to determine whether Twitter discussions regarding air pollution reflect the PM10 levels measured by official air monitoring stations in these countries. Such analyses are also performed on news article teasers, attempting to determine whether mass media portrays a realistic ambient condition and aims to promote pro-environmental behavior. The results of this study suggest that using Correlation Analysis to check for resemblance between the frequency of sentiments detected in social media discussions and the temporal changes in the PM10 concentration in the air, can serve as a measure of public awareness of air pollution. Analyzing the content of the tweets can reveal issues in the public opinion and thus, contribute to tackling them down.
Publisher:
Relation: The 19th International Conference on Informatics and Information Technologies – CIIT 2022
Identifier: oai:repository.ukim.mk:20.500.12188/25675
Identifier: http://hdl.handle.net/20.500.12188/25675
