Western Balkan societies’ awareness of air pollution. Estimations using natural language processing techniques
Journal
Ecological Informatics
Date Issued
2023
Author(s)
Madjar, Angela
Gjorshoska, Ivana
Prodanova, Jana
Dedinec, Aleksandra
Kocarev, Ljupco
Abstract
Air pollution remains a severe concern in European countries, especially in Western Balkan, where the air
monitoring data point to harmful ambient pollution. The public concern with this issue becomes particularly
critical during the fall and winter months, when the contamination is more visible, provoking a series of reactions
directed principally to the government authorities as the responsible entities for regulating air pollution levels.
Since citizen-contributed data are generally considered valuable additional information for assessing the impacts
of air pollution, the public contribution could act as a tool for increasing awareness and response about air
pollution. Consequently, this study’s objective focuses on researching public awareness of air pollution in
Western Balkan. The study assumes that citizens’ reactions will grow more intensely during the months with an
increase in air pollution levels, principally due to winter heating. Therefore, Twitter activity and news articles
related to air pollution have been investigated for the case of Macedonia, Serbia, Bosnia and Herzegovina and
Montenegro, from November 2021 to March 2022. Natural Language Processing techniques such as sentiment
analysis, topic modelling, and cross-correlations statistical analysis were employed to determine the relationship
between Twitter discussions and news with actual PM10 levels measured by official air monitoring stations. The
aim was to observe whether tweets and news teasers reflect the realistic air pollution situation. The results affirm
that social media discussions, mainly with a negative connotation, can serve as a measure of public awareness of
temporal changes in the PM10 concentration in the air and the negative consequences. The content of the resources reveals several topics of concern, contributing to better identification of public opinion and possibilities
for tracking news trends. Nevertheless, attention should be paid to news interpretation, provided that sometimes
they might offer a more neutral understanding of the situation, failing, in this way, to present the actual air
conditions and possibly impacting society in forming an unrealistic opinion. Additionally, the public might not be
able to obtain sufficient or accurate information about the primary sources of air pollution, emphasizing the need
for more transparent communication and greater education regarding air pollution monitoring. Finally, the study
provides deeper insights into the content of the data and helps detect the reasons for skepticism towards proenvironmental behavior occurring in social media discussions. Explicitly, personal disappointment with the air
quality should be taken as an inflection point by responsible parties to intervene in improving citizens’ quality of
life.
monitoring data point to harmful ambient pollution. The public concern with this issue becomes particularly
critical during the fall and winter months, when the contamination is more visible, provoking a series of reactions
directed principally to the government authorities as the responsible entities for regulating air pollution levels.
Since citizen-contributed data are generally considered valuable additional information for assessing the impacts
of air pollution, the public contribution could act as a tool for increasing awareness and response about air
pollution. Consequently, this study’s objective focuses on researching public awareness of air pollution in
Western Balkan. The study assumes that citizens’ reactions will grow more intensely during the months with an
increase in air pollution levels, principally due to winter heating. Therefore, Twitter activity and news articles
related to air pollution have been investigated for the case of Macedonia, Serbia, Bosnia and Herzegovina and
Montenegro, from November 2021 to March 2022. Natural Language Processing techniques such as sentiment
analysis, topic modelling, and cross-correlations statistical analysis were employed to determine the relationship
between Twitter discussions and news with actual PM10 levels measured by official air monitoring stations. The
aim was to observe whether tweets and news teasers reflect the realistic air pollution situation. The results affirm
that social media discussions, mainly with a negative connotation, can serve as a measure of public awareness of
temporal changes in the PM10 concentration in the air and the negative consequences. The content of the resources reveals several topics of concern, contributing to better identification of public opinion and possibilities
for tracking news trends. Nevertheless, attention should be paid to news interpretation, provided that sometimes
they might offer a more neutral understanding of the situation, failing, in this way, to present the actual air
conditions and possibly impacting society in forming an unrealistic opinion. Additionally, the public might not be
able to obtain sufficient or accurate information about the primary sources of air pollution, emphasizing the need
for more transparent communication and greater education regarding air pollution monitoring. Finally, the study
provides deeper insights into the content of the data and helps detect the reasons for skepticism towards proenvironmental behavior occurring in social media discussions. Explicitly, personal disappointment with the air
quality should be taken as an inflection point by responsible parties to intervene in improving citizens’ quality of
life.
Subjects
File(s)![Thumbnail Image]()
Loading...
Name
[7].pdf
Size
2.06 MB
Format
Adobe PDF
Checksum
(MD5):f5922cbbb1ce08fa5fb9f7a112f2a7cb