Ecological Informatics 76 (2023) 102130

Contents lists available at ScienceDirect

Ecological Informatics
journal homepage: www.elsevier.com/locate/ecolinf

Public perception of waste regulations implementation. Natural language

processing vs real GHG emission reduction modeling
Ivana Gjorshoska a, Aleksandra Dedinec a, b, *, Jana Prodanova b, Aleksandar Dedinec b,
Ljupco Kocarev a, b
a
Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Rugjer Boskovikj 16, 1000 Skopje, Macedonia
b
Macedonian Academy of Sciences and Arts, Bul. Krste Misirkov 2, 1000 Skopje, Macedonia

A R T I C L E I N F O A B S T R A C T

Keywords: The problem with waste generation and waste treatment that countries worldwide are facing, even after the
Waste management implementation of measures, raises the question of the adequacy and viability of the regulations. The waste
Waste regulations sector has been shown to contribute to a most notable increase in greenhouse gas (GHG) emissions in many
Public perception
countries worldwide, mainly coming from solid waste disposal and suggesting the need for a more profound
GHG emissions
observation of the waste management application. Therefore, this study employs an integrated approach to
Natural language processing
Sentiment analysis analyze the environmental, economic and social aspects of waste management, as the three fundamental pillars
of sustainable development. Based on the Intergovernmental Panel on Climate Change (IPCC) methodology, a
unique model was developed for the projection of GHG emissions in different scenarios, to evaluate the impact
and cost of proposed waste policies and measures adopted to the reduction of greenhouse gas emissions, as well
as the extent to which citizens themselves can contribute to the achievement of the goals. Furthermore, NLP
(Natural Language Processing) methods were used to draw conclusions about the public opinion related to the
waste sector, expressed by citizens and news media and measured on Twitter and news media teasers' data. The
results show that more than 70% of the emissions can be reduced by 2050 with a significant contribution from
the citizens, if all proposed measures are implemented, when compared to a scenario without policies and
measures. The results also suggest a strong correlation between news reports and waste-related tweets, con­
firming an alignment in public perceptions and a response connected to waste management policies. The con­
clusions of this study offer valuable implications for policy-makers, suggesting the continuous observation and
inclusion of citizens in the process of waste regulations implementation and effect.

1. Introduction 2022). The largely unsuccessful waste management is rather apparent,

With the acceleration of urban development, population and eco­
nomic growth, the amount of generated municipal solid waste (MSW) in
the world has dramatically increased, and this trend is only said to
continue in the following years. According to the World Bank, in 2020,
approximately 288 kg/capita MSW was generated globally, and the
annual waste generation is expected to increase by 73% over the next 30
years, reaching 3.4 billion tons in 2050 (Bank, 2022). Around 37% of the
global waste is disposed in some type of landfill, and 33% is mismanaged
through open dumping or burning. It is widely evidenced that inade­
quate waste management negatively impacts the environment and the
climate, public health and, ultimately, the economy (Maiurova et al.,
provided that a more adequate waste disposal is almost exclusively done
in high-income countries. As such, in the European Union, 5.2 tons of
waste were generated per EU inhabitant in 2018, 38.5% of which was
landfilled, while 37.9% was recycled (Waste Statistics - Statistics
Explained, 2022). Due to the greater economic development and EU
legislation, many of the higher recycling and collection rates happen in
Western Europe.
EU leads the way towards improving waste management by adopting
a number of directives, and by setting very ambitious goals. For
example, up to 2025, 55% of household waste should be recycled and
prepared for reuse (Directive (EU) 2018/851, 2018), and 65% of the
packaging should be recycled, while plastic bottles should be made from

* Corresponding author at: Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Rugjer Boskovikj 16, 1000 Skopje, Macedonia.
E-mail addresses: ivana.gjorshoska@students.finki.ukim.mk (I. Gjorshoska), aleksandra.kanevche@finki.ukim.mk (A. Dedinec), jprodanova@manu.edu.mk
(J. Prodanova), dedinec@manu.edu.mk (A. Dedinec), lkocarev@manu.edu.mk (L. Kocarev).

https://doi.org/10.1016/j.ecoinf.2023.102130
Received 30 December 2022; Received in revised form 13 April 2023; Accepted 14 May 2023
Available online 19 May 2023
1574-9541/© 2023 Elsevier B.V. All rights reserved.
I. Gjorshoska et al.
a minimum of 25% recycled material (Directive (EU) 2019/904, 2019).
An additional 5 % increases the targets for 2030, and for 2035 the target
for household waste is increased by another 5 %. For 2035, one addi­
tional goal is set. Namely, only 10% of the waste should end up in
landfills.
Nevertheless, the trend for enhancing collection and recycling is
expanding wider in Europe, especially to countries from the Western
Balkans, which aim to meet the EU Waste Management legislation as
part of their pro-European policies. Therefore, in the case of Serbia and
Montenegro, there were reported 338 kg and 548 kg of municipal solid
waste per inhabitant in 2019, respectively. North Macedonia reported a
consistent increase over the last ten years - from 351 kg in 2010 to 452
kg per inhabitant in 2020 (Waste Management in Montenegro, 2021;
Waste Management in North Macedonia, 2021; Waste Management in
the Republic of Serbia, 2021), unfortunately, expecting to increase lin­
early until it reaches the corresponding waste projections of EU28 in
2035 (Ministry of Environment and Physical Planning N. M., Long-term
strategy on climate action and action plan, 2021). However, already
tackling the problem of poor waste management, the latest reports in the
countries mentioned above also suggest high collection rates, between
70% and 90% of collected waste. In contrast, only up to 3% of the
collected waste is recycled, and the rest is either landfilled, in both
regulated and unregulated landfills or openly burned. In this line, the
insufficient waste treatment and increasing trend of waste generation
plays a major role in greenhouse gas emissions, where up to 80% of
emissions caused by the waste sector come from solid waste disposal in
landfills (Ministry of Environment and Physical Planning N. M., Fourth
National Communication on Climate Change, n.d.).
The waste management struggle in Western Balkans can be attrib­
uted to outdated infrastructure for collection and transport, illegal
dumpsites and the lack of facilities to sort, clean and aggregate recy­
clable materials (Waste Management in Montenegro, 2021; Waste
Management in North Macedonia, 2021; Waste Management in the
Republic of Serbia, 2021). The latest analysis (Ministry of Environment
and Physical Planning N. M., Fourth National Communication on
Climate Change, n.d.) shows that the waste sector in North Macedonia
had a 5% share in total greenhouse gas emissions in 2019. However, if
no measures are applied, it is expected to be the sector with the largest
increase in 2050 (almost four times), compared to 1990, reaching a
share of 9%. For this reason, it is essential to pay special attention to the
waste sector, emphasizing the need to reduce this trend of drastic
emissions increase, especially visible in developing countries.
Various measures are being analyzed and proposed in the literature
to reduce the greenhouse gas (GHG) emissions in the Waste sector,
ranging from recycling, composting (Paes et al., 2020), closing illegal
landfills, landfilling with landfill gas treatment, landfilling with energy
recovery and anaerobic digestion (Vázquez-Rowe et al., 2021), to
incineration (Chung et al., 2018). In order to be able to estimate the
efficacy of the waste management measures and regulations in the long
term, it is necessary to evaluate the level of GHG emissions contributed
to the waste sector, as the means for achieving carbon neutrality. An
effective way for calculating the predicted waste GHG emissions and the
policy effect has proven to be the Intergovernmental Panel on Climate
Change (IPCC) methodology, which provides procedures and guidelines
for emissions estimation for each sector from which there are GHG
emissions, including the Waste sector (Chung et al., 2018; Ramachandra
et al., 2018; Zhang et al., 2022). The governments use this methodology
to report their national greenhouse gas inventories to the United Nations
Framework Convention on Climate Change (UNFCCC).
However, inefficient allocation of external resources and insufficient
awareness among people and companies have been identified as barriers
to transitioning to a green economy (Licastro and Sergi, 2021). Citizens
have a massive role in the implementation of the measures, especially
those related to municipal waste. Public opinion and perception of the
sources of greenhouse gas emissions derived from the waste sector, as
well as the perception of how human actions affect emissions, are of
2
Ecological Informatics 76 (2023) 102130
paramount importance. There is a lack of understanding of individual
civic engagement as the vital trigger for bringing change (Teh et al.,
2022). Individuals' involvement is deemed as the central point in the
achievement of socially, economically and environmentally acceptable
consumption and waste generation, demanding a deeper insight into
consumers' emotions and perceptions of issues related to waste pro­
duction and management policies, as an instrument to better understand
their effectiveness (Oke et al., 2020).
Therefore, intending to predict the potential of waste management
measures, policies and regulations in relation to individual predisposi­
tion to contribute to their implementation, we encompass a greater
amount of data from the distinct origin (GHG inventories, individuals,
news) to integrate the environmental, economic and social aspects of
waste management. Accordingly, based on the IPCC methodology, a
model was developed for the projection of GHG emissions in different
scenarios, to show how the application of related policies and measures
can contribute to the reduction of greenhouse gas emissions in the waste
sector, while exploring the extent of citizens' contribution to the
achievement of the proposed goals. Furthermore, given that social
media have proven to represent an essential source of information on
people's perceptions, which reduces the subjectivity brought by ques­
tionnaires and interviews, the Natural Language Processing (NLP)
method was used to draw conclusions about the public opinion related
to the waste sector. For this reason, Twitter data and news media data
were collected in the period from November 1st 2021, until March 17th,
2022. This period was intentionally chosen for research, because of the
public reaction to the provision of the Law on the Management of
Packaging and Packaging Waste, referring to the restriction of single-use
plastic bags and the cost of biodegradable plastic bags, entering into
effect on December 1st, 2021. Additionally, we measured the alignment
of citizens' opinions with news content and IPCC-based environmental
and economic aspects modeling, incorporating, in this way, the three
pillars of sustainable development.
As the trend of waste generated in the world is increasing, analyzing
the Waste sector, in terms of GHG emission mitigation, has attracted
much interest in research studies. Besides, given that civic engagement
and willingness to accept policy changes are one of the main driving
wheels towards the reduction of emissions in this sector, several
research papers have evaluated public opinion and sentiments towards
waste problems (Shahidzadeh et al., 2022), (Ganczewski and Jemiel­
niak, 2022). Nonetheless, the extant literature is limited to exploring the
environmental and economic aspects of the problem, or the social
counterpart, not focusing on the waste regulations effect and their sus­
tainable management. Therefore, this research contributes to the exist­
ing body of knowledge with an integrated approach in which NLP
techniques are combined with the IPCC methodology to evaluate cor­
relations between popular news topics and public sentiments about
general waste problems, while evaluating the real GHG emission
reduction potential. Policy-makers can use the results of the research
and can contribute to the creation of new policies and measures, which
will help in the more efficient achievement of the set goals. This
approach, in essence, is universal and generalizable to all areas, where it
is imperative to underline the citizens' role in the implementation of
measures for emission reduction. It covers, for instance, energy use in
the transport and residential sectors, where individuals choose the
technology they will use, thus directly influencing the emissions from
these sectors.
As for the contribution of this research, we conclude that by using the
IPCC methodology, the uniquely developed models create different
scenarios which highlight the importance of considering all sub­
categories of the waste sector to project emissions until 2050. It em­
phasizes the need for citizen action as an essential component of
reducing greenhouse gas emissions through waste management mea­
sures. While several studies have focused on waste generation as a
process or reaction to initiatives like zero waste, recycling, and munic­
ipal waste disposal, they often neglect the role of citizens in
I. Gjorshoska et al.
implementing waste reduction measures. In that sense, this study takes a
broader perspective on the waste sector, identifying sentiments' dy­
namics concerning waste management regulations and offering insights
for policymakers on individual attitudes towards the acceptance of these
measures. By observing sentiments and news teasers before, during, and
after policies are implemented, the NLP technique proves to be a reliable
method for evaluating perceptions and feelings related to waste man­
agement regulations in a developing society in a Western Balkan
country.
The remainder of the study presents the theoretical background that
justifies the need to study the public perception of waste management
(Section 2), followed by the employed methodology (Section 3), where
we explain the process of analyzes, the data collection and the actual
content analyses. Next, we reveal the results of the estimations of the
GHG emission reduction and the public perceptions (Section 4). This
research finalizes by commenting on the findings and the conclusions
drawn from the investigation (Section 5).
2. Theoretical background
2.1. Public perception of waste management
Research of public awareness of problems related to climate change
is not uncommon, differing among sectors, observed data (social media,
questionnaires), or applied methods. Significant attention is given to the
energy sector, such as electricity generation, analyzing public awareness
of the importance of electricity generation from Renewable Energy
Sources (RES) (Almulhim, 2022) (Al-Marri et al., 2018) and the signif­
icant role of the Carbon Capture and Storage (CCS) technology (Lima
et al., 2021; Perdan et al., 2017). Notable research is presented by
(Zhang et al., 2021), where it is found that people generally have a
negative attitude towards the implementation of carbon taxation in
Europe, the USA, South Africa, Canada and Australia, concluding that
although these measures are mainly directly applicable to large pro­
duction and industrial capacities, they must all be accepted by the public
because of the fear from their impact on individuals. Similarly, (Wei
et al., 2021) explores citizens' attention and awareness regarding the
European Emissions Trading System (EU-ETS) system, underlining the
conclusion that introducing policies, especially the more severe ones,
must be harmonized with the citizens' views.
Furthermore, GHG emission reduction measures have been equally
associated with rational and sustainable consumption, where the role of
citizens is of absolute importance, in the sense of their level of awareness
about the meaning, method and benefits of rational use of resources. For
example, (Huang et al., 2022) analyze public awareness in China about
green consumption and show that environmental awareness has statis­
tical significance for household energy saving, suggesting decision-
makers to define dissemination mechanisms for offering clarified tar­
geted solutions in social media (Kuai et al., 2022). Additionally, (Aha­
mad and Ariffin, 2018) assess the level of knowledge, attitude and
practice of university students towards sustainable consumption in
Malaysia.
Similarly to responsible consumption, public opinion research
regarding the waste sector attracts attention in the literature. That oc­
curs primarily because citizens can significantly influence this sector,
probably more than any other sector, by undertaking pro-environmental
behaviors, knowledge, attitudes, and perceived control (Garcia-Vazquez
and Garcia-Ael, 2021). The waste sector understands the involvement of
different parties in the implementation of policies, processes and re­
sponsibilities to a system that encompasses waste production, selection,
treatment and disposal, with the main aim of reducing waste generation,
emphasizing the importance of waste management measures to be
accepted by the citizens (Elsaid and Aghezzaf, 2015).
Waste management research, often dedicated to promoting the
transition to a circular economy, has emphasized individual action, such
as households' waste reduction, separation and recycling, as the vital
3
Ecological Informatics 76 (2023) 102130
link for achieving environmental and economic development and sus­
tainability (Jiang et al., 2021; Kurniawan et al., 2021). The extant
literature has evidenced the importance of consumers' awareness of and
implication in the recycling chain, which, if not managed properly,
might generate more waste and endanger public health (Kumar et al.,
2018; Wu et al., 2023). In this line, various studies examine public
awareness of a particular type of waste. In (Garcia-Vazquez and Garcia-
Ael, 2021) and (Menzel et al., 2021), emphasis is placed on the aware­
ness of waste from microplastics, plastic packaging and plastic waste. It
is important to note that in Germany, for instance, there is a highly
developed awareness of the problem, and plastic-reduction in­
terventions are recommended to support consumers in acting according
to their attitudes rather than addressing only awareness and attitude
change (Menzel et al., 2021), however recalling that there are significant
geographical and cultural gaps. Furthermore, the increased usage of
electrical and electronic equipment poses a problem, not only to the
environment, but also from the social and economic aspects. (Shahrasbi
et al., 2021) reveal that in developing countries, the environmental
impact of the use of electrical and electronic equipment is of the least
importance for the public, while in developed countries, this is among
the most important aspects, when compared to the social and economic
dimensions, which calls for greater attention to increasing awareness
regarding the environmental impact of waste in developing countries.
Additionally, (Wu et al., 2021) studied the attitude of the Chinese public
towards a compulsory policy of sorting municipal solid waste. The
conclusion, somewhat surprisingly, shows a negative attitude towards
the policy, because of fines, MSW sorting rules, fees, timing of throwing
waste, and irregular recycling procedures, which could serve as a
valuable guide for policy-makers to implement measures more effi­
ciently. Finally, the food waste problem, especially during the COVID-19
crisis, has received substantial attention, especially by examining the
relationship between food waste and social media usage (Lahath et al.,
2021). In a number of studies, social media was found to have a positive
impact on raising awareness or contributing to consumer food waste
reduction (Jenkins et al., 2022). Social media have been proven as an
effective tool for promoting and facilitating activities and interaction
regarding waste management, where public perceptions, sentiments and
concerns can also be observed as a source for authorities to take action
(Jiang et al., 2021).
2.2. Social media content analysis
Social media data provides a valuable resource for evaluating public
opinion, as it allows researchers to gather large amounts of data in real-
time from a diverse range of sources, and this kind of data reduces the
subjectivity brought by questionnaires and interviews, as suggested in
(Xiao et al., 2021). They can be used to identify public attitudes towards
specific topics, understand the reasons behind these attitudes, and
evaluate the effectiveness of policies or measures aimed at changing
these attitudes (Dong and Lian, 2021). By utilizing social media data,
policymakers can gain insight into the public's views and adapt policies
to meet citizens' needs better. It can lead to more effective policymaking
and greater public trust in government institutions.
When evaluating public opinion about a particular topic using social
media data, one of the widely used techniques is content analysis, which
has demonstrated to be of substantial relevance for representing text in a
precise format (Shahrasbi et al., 2021), identifying positive or negative
trends in the public perception or revealing the reasons behind these
attitudes. Previous literature highlights the potential of social media
data for evaluating public perceptions, concluding that sentiment
analysis can be a valuable tool for policymakers to understand public
opinion and improve waste management policies (Wu et al., 2021).
In this sense, content analysis proves useful in elaborating three
components of the studied content: text preprocessing, sentiment anal­
ysis and topic modeling or clustering. Sentiment analysis is an especially
used technique in public opinion research, given its high accuracy and
I. Gjorshoska et al.
precision (Shahidzadeh et al., 2022). Studies exploring sentiments
related to waste management reveal the convenience of identifying in­
dividual observations, given the complexity of the waste management
models, especially when a transition in the system is expected (Oke
et al., 2020). Social media users' contributions act as sources of infor­
mation potentially transmitting concern, which could offer valuable
evidence regarding public views, awareness or action in compliance
with certain policies (Jiang et al., 2021).
Social media is also seen as a means for the adoption of measures or
policies and the evaluation of the effect of their implementation so as to
observe the room for change and adaptation to citizens' needs and
availability. Such an example is shown in (Pourabbasi and Shokouhyar,
2022), where a novel model for promoting mobile phone waste man­
agement is proposed, but also in (Heidbreder et al., 2021), where social
media are used for promoting plastic free July campaign. According to
(Sujata et al., 2019), social media has a substantial influence on how
individuals recycle, and this influence is fundamentally essential for all
Waste sectors.
Regarding the Macedonians' perceptions of waste management, the
authors are unaware of such research. A similar situation can be
perceived generally for the Western Balkan countries, where waste
challenges are mainly observed from the perspective of water pollution,
food-related activities or the construction industry (El Bilali, 2022;
Hassen et al., 2023; Nadazdi et al., 2022). That level of research un­
derlines the need to observe perceptions around waste management and
related regulations more thoroughly.
Fig. 1. Methodology diagram.
4
Ecological Informatics 76 (2023) 102130
3. Methods
This chapter explains the methodology used in our research in order
to cover the economic, environmental and social aspects relevant to the
evaluation of the waste sector management, as they represent the pillars
of sustainable development. A general view of the methods used is
presented in Fig. 1, and the details of each of the boxes are explained in
the following text.
3.1. Data collection, pre-processing and assumptions
3.1.1. GHG emission data and assumptions
For the purposes of projecting the development of the GHG emissions
from the waste sector, the economic (through Gross Domestic Product -
GDP) and population growth should be considered as the main drivers in
the model. Projections for these drivers for North Macedonia are made in
the Strategy for Energy Development (Ministry of Economy, N. M, n.d),
and the same values are taken for observation in this study, too. It is
assumed that the GDP growth has an average rate of 3.8% (Fig. 2), which
is a growth rate that could position a developing country, such as North
Macedonia, in 2050 closer to the levels of GDP per capita common for
today's developed Central and Eastern European countries. Additionally,
based on the UN population estimates and projections, there is a pop­
ulation decline of 0.2% (Fig. 3).
Besides these general drivers, another fundamental assumption for
projecting municipal solid waste disposal is the waste per capita.
Although, in the literature, machine learning models are also employed
for waste forecasting (Vu et al., 2022); (Adusei et al., 2022), in our
research, the projected waste per capita is based on the comparison of
I. Gjorshoska et al. Ecological Informatics 76 (2023) 102130

Fig. 2. GDP projections for North Macedonia.

(Source: Strategy for Energy Development (Ministry of Economy, N. M, n.d))

Fig. 3. Population growth for North Macedonia.

(Source: Strategy for Energy Development (Ministry of Economy, N. M, n.d))

Fig. 4. Quantity of municipal waste per capita in North Macedonia, EU28 and countries in the SEE region (in kg/capita).
(Source: Own elaboration based on data by the Eurostat)

5
I. Gjorshoska et al.
the waste per capita amount in North Macedonia with the countries in
the nearby region, as well as with the European Union 28 (EU28)
(Fig. 4). It is interesting to note that some of the countries have stable
waste per capita in the recent 10-year period, such as Austria, as well as
Greece and Croatia, while in Bulgaria, the amount of waste per capita is
reduced by about 30%. It is crucial to notice that at the EU 28 level, there
is a downward trend (presented with the black dotted line), while in
North Macedonia, there is an increasing trend (presented with the red
dotted line). It is assumed that these trends will continue until around
2030, after which there will be a decreasing trend in North Macedonia,
which would correspond to the EU 28 trend.
In the reference scenario, without measures, it is also assumed that
the composition of waste going to solid waste disposal will remain the
same during the whole period as they are for 2018, i.e., food – 36.7%,
garden – 10.7%, paper – 10.8%, wood – 0.4%, textile – 3.7%, nappies –
5.0%, plastic and other inert – 32.6%. Additionally, the distribution of
waste-by-waste management treatment will be equal to the distribution
in 2018, for the whole period. For the industrial waste calculations, the
data from the Strategy for Energy Development for the production index
growth for the industry was used.
3.1.2. Twitter data
For the purpose of the study, we collected tweets related to waste
content using Twitter's stream Application Programming Interface (API)
with an authenticated application, which connects to a public stream
consisting of a sample of tweets posted on Twitter (Roesslein, 2020).
Although sentiment analysis research often relies on using hashtags for
filtering tweets, the selection of relevant tweets in our case was made
using keywords instead of hashtags. This decision was made due to a
lack of hashtags relevant to this topic in tweets from the Western Balkan
region. Macedonian and Serbo-Croatian tweets were filtered with waste-
related keywords, such as “отпад“(waste), “otpad“(waste),
“ѓубре“(garbage) and “рециклира“(recycle). Waste-related tweets were
collected on a weekly basis over a period of 4 and a half months starting
from November 1st, 2021, until March 17th, 2022.
This period was crucial since, in the second half of 2021, there have
been significant changes regarding waste legislation in the country.
Namely, besides the more general Law on Environment, the Government
of North Macedonia has enacted more specific laws on waste manage­
ment, management of packaging and packaging waste, the re­
sponsibility of producers, management of additional waste flows,
electrical and electronic equipment and waste electrical and electronic
equipment, batteries and accumulators and waste batteries and accu­
mulators. Among these regulations, the most drastic measure concern­
ing the public was the Law on the Management of Packaging and
Packaging Waste, which refers to single-use plastic bags and the cost of
biodegradable plastic bags, put into effect on December 1st, 2021. In the
expectance of this law entering into force, we decided to consider the
public sentiments about this problem as the main focus of this research.
After the initial collection of the tweets, they were saved as an Excel
spreadsheet for further data cleansing, which included checking and
filtering the ones associated with the waste-related topics. This sec­
ondary filtering step was necessary for this study due to language am­
biguity. Namely, words such as “trash” and “garbage” tend to be used as
derogatory terms for people, objects or situations not relevant to this
study objective. The collected tweets were translated into English using
an online document translation tool, and parts of the tweets, such as
“RT”, standing for retweet and usernames, were removed from the
tweets in order to focus on the tweets' content (Online Doc Translator,
2021). The translated text is manually rechecked.
3.1.3. News data
Journalism in Western Balkans significantly enhances societal tran­
sitions (Andresen et al., 2017), where pollution issues are constantly
covered and evaluated. The decision to analyze news reports was made
with the idea of comparing the alignment of the news media content
6
Ecological Informatics 76 (2023) 102130
with the public sentiment regarding waste topics. For the data collec­
tion, we used a web scraping method by the Beautiful Soup python li­
brary to scrape the news reports from time.mk and time.rs websites,
which are cluster-based news websites that collect and aggregate news
from 120 distinct sources (Richardson, 2007; Time.mk, 2022; Time.rs,
2022). Similarly, as the tweets, the news reports were filtered using the
keywords “otpad” (waste) and “отпад” (waste) and were collected
weekly over the same time period as the Twitter data. During this pro­
cess, we collected both the news title and the news preview to get a
better picture of the contents of the news reports for the respective week.
The news reports datasets were stored, translated and manually checked
using the same techniques as the waste-related tweets.
3.2. Method for evaluation of GHG emissions
For the evaluation of the GHG emissions from the waste sector up to
2050, a completely new model was developed, based on the method­
ology implemented in the IPCC software. This model covers all the
subcategories from the waste sector, and instead of calculating only the
historical emissions, it is able to project the emissions in the period up to
2050, by varying the activity data and inserting different assumptions,
thus enabling the creation of different scenarios.
For the purpose of this study, two different scenarios are presented.
The first is without measures – WOM scenario, and the second one is
with all measures (WAM) scenario, in which all the proposed measures
(presented in the following section) are implemented. The WOM sce­
nario is used for comparison to estimate the emissions coming from the
waste sector if no policies or measures are implemented. In contrast, the
mitigating WAM scenario shows the potential of emission reductions up
to 2050, if corresponding measures are applied.
For projecting the emissions for some of the categories in the WOM
scenario, simple linear regression is used, based on the historical data.
For example, for the emission from mechanical and biological treatment
with composting, an equation for the trendline of the emissions from
composting is obtained. Based on this equation, the emissions for the
period from 2020 to 2050 are calculated (Fig. 5).
The emissions from waste incineration are also considered in the
WOM scenario. Again, a trendline is calculated based on the available
historical data for the period 2000–2019 (Fig. 6) to calculate the emis­
sions from waste incineration up to 2050.
In order to estimate the emissions from the industrial wastewater
sector, a correlation is made between the total organic degradable ma­
terial in wastewater with the value added in the industry from 2014 to
2018 (Fig. 7). The derived equation for the correlation is used to
calculate the total organic degradable material in wastewater for the
period up to 2050, which is used to project the amount of emissions from
industrial wastewater (Fig. 8).
Furthermore, in order to be able to evaluate the impact of waste
policies and measures for reducing GHG emissions, we address five
policies/measures which encompass the overall Waste sector. These
policies/measures are in line with the various strategic documents, such
as the National Waste Management Plan 2021–2031, Strategy for Waste
Management in the Republic of Macedonia, Regional Waste Manage­
ment, National Communication, Biennial Update Reports and Long-term
strategy on Climate Change:
- Landfill gas flaring – assumes opening new regional landfills and
rehabilitation of the existing landfills and illegal (“wild”) dumpsites
with very high, high and medium risk in each of the eight waste
management regions. The rehabilitation includes covering the
existing non-compliant landfills, supplemented by gas extraction and
flaring. Table 1 presents the assumptions for the timeline for opening
the regional waste management regions.
- Mechanical and biological treatment (MBT) in new landfills with
composting –an installation of a system for MBT and composting is
proposed. It includes the reduction of the amount of not treated
I. Gjorshoska et al. Ecological Informatics 76 (2023) 102130

Fig. 5. Calculation of waste treated by biological treatment facilities.

(Source: Own elaboration based on data by the Ministry of Environment and Physical Planning, Total annual report of processed data on the quality of the
environment)

Fig. 6. Calculation of the total amount of waste incinerated.

(Source: Own elaboration based on data by the Ministry of Environment and Physical Planning, Total annual report of processed data on the quality of the
environment)

- Selection of waste –installation of containers for collection of

- Improved waste and materials management at industrial facilities –
- Improved industrial wastewater treatment – takes into account in­
Fig. 7. Correlation between the total organic degradable material in waste­
water and production index growth in the industry for the period 2014–2018.
biodegradable components (garden and food waste) of municipal
waste, expressed as a reduction of the percentage of biodegradable
components in generated municipal waste.
selected waste, mainly paper, but also wood, textile, nappies, plastic
and other inert waste is assumed. This will help to achieve the goals
for recycling municipal solid waste of 25% in 2025, 45% in 2035 and
65% in 2045, as defined in the National Waste Management Plan
2021–2031. The specific goals for recycling each type of waste are
presented in Table 1.
takes into account the Extended Producer Responsibility schemes,
which enables and contributes to achieving the goals of the circular
economy. As shown in Table 1, it is assumed that the percentage of
industrial waste treatment will increase according to the goals
defined in the Law on Management of Packaging and Packaging
Waste, for each type of industrial waste.
dustrial sludge treatment processes which include mechanisms and
processes used to treat waters that have been contaminated in some
way by anthropogenic industrial or commercial activities prior to its
release into the environment or its re-use. The measure regarding the
treatment of industrial wastewater is also part of the integrated
environmental permits of the industrial facilities. Table 1 shows the
assumed percentage of emission reductions in 2050 by implementing
this measure.

7
I. Gjorshoska et al.
Fig. 8. Total organic degradable material in wastewater and production index growth in the industry for the period 2012–2050.
(Source: Own elaboration based on data by the State Statistical Office, North Macedonia (for Industrial Index Growth))
3.3. Methods for evaluation of public perception
The most basic use of content analysis with NLP is text preprocessing,
as the method for representing the text in a format suitable for com­
puters (Shahrasbi et al., 2021), i.e. in vector format, which can be done
using, for example, simple bag-of-words representation (Loureiro and
Alló, 2020), transformation using Term Frequency - Inverse Document
Frequency (TF-IDF) (Loureiro and Alló, 2020) or more advanced options
such as sentence-BERT framework, which achieves state-of-the-art per­
formance on various problems (Grootendorst, 2022). Furthermore, to
extract relevant information in terms of the positive or negative attitude
represented in the analyzed content, sentiment analysis can be
employed, based on machine learning, such as in the case of (Huang
et al., 2022), where a combination of Convolutional Neural Networks
(CNN) and Long Short-Term Memory (LSTM) networks is used for this
purpose or on lexicon dictionary, such as Emolex (Loureiro and Alló,
2020), VADER (Hutto and Gilbert, 2014), SOCAL (Taboada et al., 2011).
Furthermore, techniques for topic modeling, such as Latent Dirichlet
Allocation (LDA) or variations of it (Huang et al., 2022) (Wei et al.,
2021) or more optimized algorithms developed for short texts, such as
GPU-PDMM (Shahrasbi et al., 2021), can be useful in the identification
of motives for the recognized positive or negative trends in the senti­
ment analysis.
A variety of research has resorted to general clustering methods to
discover the essential topics that represent the documents, such as, for
example, the bisecting k-means algorithm used to cluster the tweets
sample, and text summarization techniques identify the topic of each
cluster (Zhang et al., 2021). However, a method that has attracted much
attention in the literature due to its competitive and stable performance
across various tasks (Grootendorst, 2022) is the BERTopic, which is
represented by a clustering model – hierarchical DBSCAN, but the topic
of each cluster is calculated using a class-based TF-IDF method.
3.3.1. Sentiment analysis
Sentiment analysis is the practice of using algorithms to classify
various samples of related text into positive and negative categories.
This method is widely applied in analyzing customer reviews and re­
sponses and for analyzing online opinions on online social media plat­
forms. Sentiment analysis within microblogging has verified Twitter as a
valid online indicator of political sentiment reflecting the offline polit­
ical landscape (Tumasjan et al., 2010), social events or healthcare issues
(Gohil et al., 2018; Zhou et al., 2013).
Several models and tools can be used for the sentiment analysis of a
text. However, none of the tools available right now can perform
sentiment analysis on Macedonian or Serbo-Croatian text, resulting in
the previously explained English translation of the collected data,
8
Ecological Informatics 76 (2023) 102130
described in Section 3.1.2. For the purpose of the sentiment analysis in
our research, we used a prebuilt Sentiment Intensity Analyzer (Bird
et al., 2009; Hutto and Gilbert, 2014). The analyzer is built using
VADER, a human-centered sentiment lexicon that combines lexical
features from well-established sentiment word-banks and features that
are common in microblogging, including acronyms, slang and emoji.
The reasons for choosing VADER are twofold: (i) the fact it is an open-
source tool, allowing easy replication of the procedures we performed
in our study; and (ii) VADER's expressive results observed in previous
experiments. Each candidate feature of this lexicon has been assessed
using a Wisdom-of-the Crowd approach, as well as using identified
heuristics humans use to assess sentiment intensity of text, which leads
to creating a human-validated sentiment intensity on corpora within
four domains including social media text, movie reviews, technical
product reviews and opinion news articles. The Sentiment Intensity
Analyzer from the NLTK library uses this lexicon to assess the sentiment
intensity on a piece of text by verifying the availability of the words
composing the content in the lexicon and produces measurements that
assess the sentiment intensity in the text from the identified words.
Using this tool, the tweets and news reports for each week were classi­
fied into three groups, i.e., tweets and news reports containing positive,
negative or neutral sentiment.
3.3.2. Dynamic topic modeling
Topic modeling is a method for the unsupervised classification of
documents, which finds natural groups of items within a collection and
is used for discovering hidden semantic structures in a body of text.
Within our analysis, we used a BERTopic model that leverages BERT
embedding and c-TFIDF to create dense clusters allowing for easily
interpretable topics (Grootendorst, 2022). BERTopic uses document
embedding representation with a pre-trained Sentence-BERT frame­
work, which allows converting sentences and paragraphs into vector
representations using the well-known BERT language model. This sen­
tence embedding is then clustered using Hierarchical DBSCAN, an
extension of the DBSCAN algorithm that is able to find clusters of
varying density, and Uniform Manifold Approximation and Projection
(UMAP) for vector dimensionality reduction.
Lastly, topic representations are extracted from the clusters using
class-based TF-IDF. Furthermore, to create dynamic topic modeling,
BERTopic is first fitted on the entire corpus, not taking into account the
temporal aspect, in order to create a global view of topics. Then, the
local representation of each topic is done using the following equation:
( )
A
Wt,c,i = tf t,c,i *log 1 +
tf t
where tft, c, i is term frequency which models the frequency of the term t
I. Gjorshoska et al. Ecological Informatics 76 (2023) 102130

Table 1
Proposed policies/measures in the Waste sector.

Sector Budget Source of finance Assumptions

1. Landfill gas flaring

Closing of existing and opening of new landfills by waste management regions in the
following order:

• Skopje – 2025
Solid waste 10.5 mil. Local self-government through Public Utilities, Public
• East and Northeast – 2026
disposal EUR Private Partnership, EU funds, IFIs
• Polog – 2027
• Southeast – 2030
• Pelagonia and Southeast – 2030
• Vardar 2030

2. Mechanical and biological treatment (MBT) in new landfills with composting

Opening of the regional landfills in the following order:

• Skopje – 2025
• East and Northeast – 2026
• Polog – 2027
• Southeast – 2030
Solid waste 36.1 mil. Local self-government through Public Utilities, Public
• Pelagonia and Southeast – 2030
disposal EUR Private Partnership, EU funds
• Vardar 2030
Reduction of the amount of not treated biodegradable components of municipal waste
on an annual level, expressed as a reduction of the percentage of biodegradable
components in municipal waste generated, is assumed with a rate as defined in Table 8
of the National Waste Management Plan 2021–2031 (Ministry of Environment and
Physical Planning, 2021).

3. Selection of waste
Gradual increase (compared to WOM) of

• paper selection reaching 82% in 2050

Solid waste 2 mil. Local self-government through Public Utilities, Public
• wood selection reaching 36% in 2050
disposal EUR Private Partnership, EU funds
• textile selection reaching 32% in 2050
• nappies selection reaching 50% in 2050
• plastic and other inert waste reaching 69% in 2050

4. Improved waste and materials management at industrial facilities

Ministry of Environment and Physical Planning, The percentage of industrial waste treatment will increase according to the goals
Solid waste
n/a Municipalities and city of Skopje, Industrial facilities, EU defined in the Law on Management of Packaging and Packaging Waste, for each type of
disposal
funds industrial waste.

5. Improved industrial wastewater treatment

Industrial Ministry of Environment and Physical Planning, The emissions from industrial wastewater will be reduced by 50% in 2050 when
n/a
wastewater Industrial facilities, EU funds compared to the emissions from this category in 2050 in the WOM scenario.

Source: Own elaboration based on the latest national strategic documents, such as the National Communication, Biennial Update Reports and Long-term strategy on
Climate Change.

in a class c, at time i. The “classical” inverse document frequency term is
modified and, in this case, represents a metric of how much information
a term provides to a class. It is calculated by taking the logarithm of the
average number of words per class - A divided by the term t frequency
across all classes.
For the purpose of our analysis, fine-tuning on a pre-trained BER­
Topic model was done using a minimum topic range of 50 and mono­
grams, bigrams and trigrams, which were derived from the translated
tweets and news reports during the fine-tuning process. Additionally,
because the BERTopic model is used to extract topics from large docu­
ments, a count vectorizer model was used to ensure relevant topic
extraction (Pedregoa et al., 2011). The topic over time modeling was
done using the fine-tuned BERTopic model with weekly timestamps and
global and evolution tuning.
3.3.3. Bigram similarity
In order to evaluate the similarity between the tweets and news re­
ports, we used soft cosine similarity as a method of assessing semantic
similarity between two documents. This type of measure considers two
vectors as pairs of features and considers their similarity in a vector
space model (Sidorov et al., 2014). Given two N-dimension vectors a and
b, the soft cosine similarity is calculated as follows:
∑
N
si,j ai bj
i,j
soft cosine1 (a, b) = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑N ∑N
si,j ai aj * si,j bi bj
i,j i,j
where sij = similarity (featurei, featurej).
To measure the soft cosine similarity between the most common
terms found in tweets and news reports for each week, the top 10 most
common bigrams for tweets and news reports were determined using a
count vectorizer model (Pedregoa et al., 2011). All bigrams taken from
tweets and news reports were respectively sorted in ascending order,
and a similarity matrix was calculated using GloVe vector representa­
tions (Pennington et al., 2014). Soft cosine similarity was calculated on
the respective vector representations of the bigrams for each week.

9
I. Gjorshoska et al.
4. Results
4.1. GHG emission reduction estimation
The results from the IPCC modeling generally show that if no policies
or measures are taken in the waste sector, the emissions will increase by
137% in 2050 when compared to the 2019 level (Fig. 9), or they will be
4 times higher when compared to 1990 level (Fig. 10). Regarding the
categories, solid waste disposal will remain most dominant, but followed
by industrial wastewater in the WOM scenario in 2050.
It can be noted that in the case of the WAM scenario, instead of an
increase, these measures can reduce emissions by 32% in 2050
compared to 2019, meaning that they will be almost at the same level as
in 1990 (Fig. 11). If it is compared to the WOM scenario, over 70% lower
emissions will be obtained in 2050 (Fig. 10).
In addition to the analysis of the WAM scenario, the potential for
reducing GHG emissions was evaluated for each policy and measured
separately (Fig. 10). The results show that the highest GHG emission
reduction potential is in covering the existing wild dumpsites and
opening regional landfills in 2025–2030, as expected. The effects of this
measure can be seen immediately in the drastic reduction of emissions,
as new regional landfills are opened, and old ones are closed. In contrast,
with the implementation of the other measures, an increasing trend can
be observed because the emissions in a particular year are based on the
accumulated waste from the previous years. Thus, the emission reduc­
tion effect is even more visible in the period after 2050. However, the
increase in the period 2020–2050 for each of the measures is at a much
slower rate when compared to a WOM scenario.
Regarding the industry, by implementing the Extended Producer
Responsibility schemes to improve waste and materials management at
industrial facilities towards a circular economy, the emissions can be
reduced by 6% in 2050 when compared to the WOM scenario. Proper
treatment of industrial wastewater is also critical since, if no measures
are taken in this sector, the emissions will increase ten times in 2050,
when compared to 2019, provided that they are correlated to the in­
crease of the GDP of the country. However, if mechanisms and processes
are used to treat contaminated industry waters, emissions can be
reduced by 25% in 2050, compared to the 2019 level.
It is interesting to see the potential of reducing emissions in the
measures in which people, that is, their awareness, has a significant part.
These are the measures for waste selection, i.e. for the selection of paper,
wood, textile, nappies, plastic and other inert waste and the MBT mea­
sure, where the selection of biodegradable components (garden and food
waste) of municipal waste is required. It is important to note that the
waste selection measure can reduce emissions by about 9% in 2050,
compared to the WOM scenario, while the MBT measure can reduce
them by about 14%. For the implementation of these two measures, the
human factor is of particular importance and it has a primary role in
their fulfillment. In that direction, it is necessary to evaluate the level of
awareness of people in the country about the emissions deriving from
Fig. 9. Total GHG emissions by categories from the Waste sector – WOM scenario (in Gg CO2-eq).
10
Ecological Informatics 76 (2023) 102130
the waste sector and, if necessary, to propose appropriate initiatives and
additional policies aimed at raising people's awareness about their part
in emissions mitigation.
When it comes to how the proposed measures affect the categories
from the waste sector, it can be noted that by MBT, the emissions for
biological treatment will be increased when compared to the WOM
scenario, but at the same time, the solid waste disposal emissions will be
reduced. The emissions from domestic wastewater and open burning in
2050 will decrease mainly as a result of the decrease in the population
and the waste per capita, when compared to the 2019 level. Besides the
measure proposed for treating industrial wastewater, this sector has a
dominant role in the 2050 WAM scenario.
Most of the GHG emissions from the Waste sector are emissions of
CH4, but after 2025 the CO2 emissions start to increase due to the
introduction of the process of flaring in the WAM scenario (Fig. 12).
Although the share of the CO2 emissions increase, the total GHG emis­
sions will be drastically reduced, by implementing this measure.
4.2. Public perception
Regarding the evaluation of the public perception of the emissions
from the waste sector, Figs. 13 and 14 show statistics for the collected
data over the analyzed period. The sentiment analysis of the data from
both sources shows predominantly negative sentiment over positive and
neutral content. Additionally, the percentage of texts classified as
negative is much higher in the news reports compared to the percentage
of negative sentiment in tweets. In Fig. 14, a stable trend of news reports
over time can be detected, however, the trend of collected tweets shows
a few bigger peaks. Namely, the first peak around the beginning of
December 2021 is connected to the provision of the Law on the Man­
agement of Packaging and Packaging Waste and, particularly, the ban on
single-use plastic bags and the additional rise of the cost of biodegrad­
able plastic bags from 2 Macedonian denars to 15 Macedonian denars
(Georgievski, 2021). The second peak appearing around the end of
January 2022 corresponds to the news that 300.000 expired COVID-19
vaccines would be landfilled by the end of February due to a lack of
interest in COVID-19 vaccination. The third bigger peak can be noticed
towards the end of February and beginning of March 2022, which is
connected to the war conflict between Russia and Ukraine, precisely the
Russian occupation of Chernobyl and possible risks of major nuclear
disaster and radioactive contamination across Europe (Polityuk and
Crellin, 2022). Other smaller peaks include topics associated with plastic
waste generation during the COVID-19 pandemic (between November
15th and November 29th), total generated municipal waste (between
the end of December 2021 and the beginning of January 2022), as well
as electronic waste and landfills regulations (around January 16th,
2022).
The sentiment analysis of the collected tweets and retweets is shown
in Fig. 15 and Fig. 16. The percentage of tweets with negative sentiment
is much higher compared to the percentage of positive and neutral
I. Gjorshoska et al. Ecological Informatics 76 (2023) 102130

Fig. 10. Emission reduction potential by each proposed measure.

Fig. 11. Total GHG emissions by categories from the Waste sector – WAM scenario (in Gg CO2-eq).

Fig. 12. Total GHG emissions in the Waste sector by gasses (in Gg CO2-eq).

11
I. Gjorshoska et al.
Fig. 13. Statistics about tweets/news.
Fig. 14. Number of tweets and news reports over time.
Fig. 15. Sentiment trends in tweets (%).
Fig. 16. Sentiment trends in retweets (%).
12
Ecological Informatics 76 (2023) 102130
Fig. 17. Trends in number of news.
tweets over the analyzed period. A more considerable percentage of
negative tweets can be observed in mid-November 2021 when the ma­
jority is related to the ban on single-use plastic bags, provided that this
was the period when the news about the Law on the Management of
Packaging and Packaging waste first started being reported in the news.
Additionally, this period is associated with the mismanagement of bulky
waste, usage of nuclear power for power supply and amounts of waste
generated during the COVID-19 pandemic. Other peaks can be observed
at the beginning of January 2022, related to the reports of generated
waste during the pandemic worldwide, and towards the end of February
2022, when the predominant topic is the Russian-Ukrainian war. It is
interesting to note that during the beginning of December 2021, when
the Law on the Management of Packaging and Packaging waste took
effect, there is a decrease in the percentage of tweets with negative
sentiment and an increase in positive tweets (Fig. 15). The trends in
sentiment for the retweets during the analysis period also suggest that
most of the time, the negative tweets are being shared, however, there is
a few weeks period where a bigger percentage of positive retweets can
be observed. These peaks are in accordance with the topics of nuclear
waste for power supply, the Law for package management and initia­
tives for selecting waste and cleaning up landfills.
Figs. 17 and 18 show the trends in the news reports during the period
from November 1st 2021 until March 17th 2022. In Fig. 17, it is evident
that there are several periods when more news reports have been pub­
lished, which are greatly associated with the events observed during the
analysis of the Twitter data. Additionally, both Fig. 17 and Fig. 18
suggest that there is a much higher number of negative news published
during the analyzed period. When comparing the sentiment of the
content from Twitter and news reports, it can be observed that there are
very similar trends in the percentage of the sentiments. That is especially
evident in the trends regarding negative posts and news reports.
The dynamic topic modeling analysis also shows similarities between
the tweets and the news reports, as displayed in Figs. 19 and 20. Fig. 19
shows the trends of the 10 most common topics identified among the
tweets. Some of these topics include: people inadequately throwing
waste, plastic bags, nuclear waste as a result of nuclear power plants,
medical waste and its management by the municipalities, burning wood
and other materials for the purpose of heating, disposal of electronic
waste such as lithium batteries, composting bio waste and vehicle
recycling. All of the identified topics have similar frequency rates, with
the exception of the topic of plastic bags, which is connected to North
Macedonia's ban on single-use plastic bags (Georgievski, 2021), which is
concentrated near the date when the relative regulation entered into
force. Fig. 20 shows the trends of the most common topics identified in
news reports. There are fewer identified topics within the news reports
data compared to the Twitter data. The topics include: waste landfills
and waste management in Skopje, the “plastic waste pandemic”, medical
waste and its management by the municipalities, plastics bags for
I. Gjorshoska et al. Ecological Informatics 76 (2023) 102130

100 2014). The highest values for cosine similarity, while not taking into
90 consideration the sentiment, are 0.86, 0.81 and 0.79, corresponding to
80 the periods of the beginning of December 2021, the beginning and end of
70 February 2022, and mid-March 2022. The periods referring to the
60 negative beginning of December and during February are in accordance with the
50 positive previously mentioned news about the plastic bags ban, vaccine disposal
40 neutral and war conflict between Russia and Ukraine. In contrast, the mid-
30 March period also suggests a high similarity value (0.81) and is con­
20 nected to various news reports such as inadequate waste disposal, low
10 recycling and waste selection rates and initiatives for landfills cleaning.
0 Similar scores can be observed for cosine similarity of topics with
10/31/2021

12/12/2021
12/19/2021
12/26/2021
11/7/2021

12/5/2021

1/16/2022
1/23/2022
1/30/2022

2/13/2022

2/27/2022
11/14/2021
11/21/2021
11/28/2021

1/2/2022
1/9/2022

2/6/2022

2/20/2022

3/13/2022
3/6/2022
negative sentiments, where the corresponding cosine similarity scores
for the three major news events are 0.84, 0.83, 0.82 and 0.78.

Fig. 18. Sentiment trends in news (%). 5. Discussion

5.1. Theoretical conclusions

packaging and space debris. The trends for the previously mentioned
topics show bigger peaks during the beginning of December and towards
For estimating the GHG emission mitigation possibilities in the waste
the end of January, corresponding to the two aforementioned news
sector, the IPCC methodology was implemented to offer different sce­
events about the plastic bags ban and vaccine disposal.
narios for the case where measures are taken in the waste sector (WAM)
Lastly, Fig. 21 shows the semantic similarity between the most
and where measures are not taken (WOM). The results from the poten­
common bigrams found in tweets and news reports over time. The figure
tial for GHG emission mitigation from the waste sector show that more
shows trends for bigram similarity for all collected data for each week, as
than 70% of the emissions can be reduced in 2050 if all proposed
well as the similarity between bigrams of tweets and news reports
measures are implemented, when compared to the WOM scenario,
classified as positive or negative accordingly. The results reveal a strong
keeping the emission from this sector at almost the same level as in
semantic similarity between the most common topics over time, since
1990. The significant potential for reducing emissions from the Waste
the threshold for strong soft cosine similarity is 0.5 (Sidorov et al.,
sector, with the application of appropriate measures, is also evident in

Fig. 19. Topics over time - Tweets.

Fig. 20. Topics over time - News reports.

13
I. Gjorshoska et al.
Fig. 21. Soft cosine similarity between tweets and news reports.
the literature, such as for Korea, where certain measures can achieve a
17% reduction in emissions in 2050 compared to 2010 from the Waste
sector. Additionally, China investigates the possibility of climate
neutrality of the Municipal Solid Waste sectors, as it plans to achieve
overall carbon neutrality before 2060. Along with state institutions and
industry, the citizens themselves have a considerable share in the
implementation and achievement of such results, especially in the Solid
Waste Disposal sector.
Moreover, in this study, an analysis was made of the online news
media reports about waste, as one of the primary sources from which
citizens obtain pollution information, in general, regardless of the
source. The results were contrasted with the findings from the sentiment
analysis of Twitter data, revealing the accordance between news media
and public sentiment content. In that sense, besides the environmental
and economic aspects of the waste management regulations, this study
tackles the social aspect, too, by evaluating public awareness and un­
derstanding of the proposed measures.
The results obtained from all three analytical methods underline that
a greater consciousness around reducing the amount of waste, reuse of
resources and selection of waste is mainly required. The strong corre­
lation between the news reports and waste-related tweets suggests that
news media content is highly aligned with the overall public response to
waste management policies. Three major events were identified as in­
citements of a larger public response during the analyzed period.
The first event, the one that established the objective of this study,
was the law regulating very lightweight plastic bag use in North
Macedonia and the increase of biodegradable plastic bags' prices,
enforced as of December 1st, 2021 (Georgievski, 2021). The ultimate
goal of these legal changes was to promote a switch towards a more pro-
environmental action, by increasing the consumption of reusable bags
made of eco-friendly materials. The money paid for purchasing more
expensive bags goes to a special “environmental fund” at the Ministry of
the Environment and Physical Planning, and only in December 235,000
EUR were collected. According to the Ministry, this budget will be spent
to invest in more ecological projects (Од пластични кеси собрани 235
илјади евра, 2022).
However, even though this measure complies with EU regulations,
the results show negative sentiment across both data sources, news
media and Twitter data. The main reason for this result is probably due
to people's skepticism towards the governing bodies responsible for
waste management, as it has already been identified in previous
research of similar regulations' acceptance (Little et al., 2019). Addi­
tionally, the vast majority of Twitter users during this period com­
mented on the ineffectiveness of the ban because of the common
14
Ecological Informatics 76 (2023) 102130
practice of disposing of mixed household waste in single-use plastic
bags. The topic analysis showed that issues such as low recycling and
waste selection rates, as well as active burning of toxic waste, were
discussed during this event. Therefore, although the capacity of the
competent institutions to implement the measure is criticized, the real
reason for the introduction of a higher price for biodegradable bags,
intended to raise people's awareness about the use of more ecological
materials and reusable bags, is rarely seen in people's attitudes. There­
fore, it can be concluded that although adopting such legislation is under
waste management policies and measures expected to mitigate GHG
emissions, it is followed by mainly negative remarks from both news
media and social media. It is noteworthy to highlight that this finding is
not unexpected given the numerous studies that have found public op­
position to the adoption of laws and other actions that have a positive
impact on the environment (Shahrasbi et al., 2021; Wu et al., 2021;
Zhang et al., 2021). However, this conclusion suggests that, in order for
the measures to be successfully implemented, it is necessary to
concentrate on implementing additional measures for increasing public
knowledge, reaching a level of awareness similar to the one presented in
(Menzel et al., 2021). In this sense, the findings could be considered for
implementing similar regulations related to pro-environmental
behavior, and not just in terms of exploring possibilities for educating
the public but in creating a pro-active response to the climate change
consequences.
The second event that posed a significant initiator of public reaction
was the great scale of expired COVID-19 vaccine disposal. In this
instance, the majority of Twitter users showed concerns about the
management of medical waste, not only for the event of vaccine
disposal, but also for the increased amount of medical waste generated
over the last years as a result of the COVID-19 pandemic (Tonnes of
COVID-19 Health Care Waste Expose Urgent Need to Improve Waste Man­
agement Systems, 2022). The Twitter users' responses to 9 are centered
around the distrust of the waste managing facilities in the ability to
adequately treat medical waste, since poor administration of this type of
waste can greatly threaten human and environmental health. Such
concerns might be justified and expected, bearing in mind previous in­
cidents of inappropriate medical waste treatment by certain companies,
although in this case, the vaccines were destroyed at the Drisla landfill,
the only managed landfill in North Macedonia. Our results contribute to
the research analyzing the effect of medical waste created during the
COVID-19 pandemic (Harm, 2021; Rayhan et al., 2022) and, likewise,
emphasize the inappropriate management of assets found indispensable
for public health during the COVID-19 crisis. The exact footprint that
this additional medical waste poses on the GHG emissions in North
I. Gjorshoska et al.
Macedonia is still not evaluated since the last year for which GHG in­
ventory was prepared is 2019. Nevertheless, although medical waste
incineration accounted for only 0.1% of total greenhouse gas emissions
in 2019, this is not an issue that should be neglected, given that it speaks
about the poor management of resources. Especially relevant because
they regulate public health and related consequences and risks, health
policies and interventions' further research should consider all possible
sources of pollution affecting citizens' well-being (Gollust et al., 2017).
The last identified event provoking a visible reaction among the
public was the armed conflict in Ukraine. The Russian invasion of
Ukraine began on February 24th, 2022, and shortly after, the towns of
Chernobyl and Pripyat were under Russian control. The primary concern
about the nuclear disaster happened a few days after the initial invasion,
following the reports that the radiation level in Chernobyl drastically
increased due to the failure to maintain the nuclear reactor facility
(Polityuk and Crellin, 2022). The reaction of the Macedonian users of
Twitter was following these concerns, especially because of the
geographical proximity of the affected region. Although these concerns
show public awareness of the adverse effects of the harmful levels of
radiation from uranium, this is not correlated to GHG emissions and thus
is not particularly relevant for this study.
5.2. Implications
The continuous consumeristic trend worldwide has led to a great
amount of waste generation, challenging to be treated, especially when
it comes to developing countries with limited resources and regulations.
In order to avoid irrevocable consequences, current research calls for
practical regulations that would offer socially, economically and
ecologically reasonable solutions headed towards pro-environmental
behaviors (Oke et al., 2020). Accordingly, it is necessary to accurately
predict waste generation and the provoked GHG emission so that
adequate waste management can be modeled and more pragmatic pol­
icies and measures can be planned (Adusei et al., 2022; Kittipongvises
et al., 2022). In this sense, the IPCC methodology is a widely accepted
approach that has proven to be valuable not only for reporting historical
emissions but also for projecting and forecasting emissions in the future
and evaluating the effect of implementing related policies or measures.
That has been demonstrated in this paper and supported by previous
research conducted for evaluating the mitigation pathways up to 2050
(Chung et al., 2018) and identifying options for achieving carbon
neutrality in specific waste sectors (Zhang et al., 2022).
With this in mind, the general goal of this study is to predict GHG
emissions trends concerning waste management regulations imple­
mentation but also to capture public opinion regarding waste manage­
ment policies, understand sentiments and recognize concerns around
waste management topics. The uniquely developed models, using the
IPCC methodology, create different scenarios that emphasize the need to
consider all the subcategories of the waste sector for projecting emis­
sions up to 2050. In this line, underlining the importance of citizens'
actions as one of the essential wheels in the run for GHG emission
mitigation through implementing waste measures distinctively adds to
the waste management body of knowledge.
Moreover, a variety of studies have focused their interest on waste
generation as a general process or a reaction to a specific initiative, such
as zero waste trend, recycling, and municipal waste disposal (Adusei
et al., 2022; Ganczewski and Jemielniak, 2022; Kumar et al., 2018),
rather than looking into citizens as one of the main actors in the
implementation of the related measures for achieving waste generation
decrease. This study observes the waste sector from a broader perspec­
tive, identifying sentiments' dynamics in accordance with waste man­
agement regulations and offering insights for policy-makers into the
individual propensity towards measurements' acceptance and imple­
mentation. The fact that sentiments and news teasers are observed
before, during and after policies are enacted provides the advantage of
more objectively evaluating the public awareness of waste issues. With
15
Ecological Informatics 76 (2023) 102130
all this, the NLP technique of sentiment analysis and topic modeling
proves to be a reliable solution for evaluating public perceptions and
feelings regarding waste management regulations, taking effect in a
developing society of a Western Balkan country.
Additionally, as a suitable tool to reduce data subjectivity (Huang
et al., 2022), social media have proven to be an essential data source for
extracting large amounts of information from consumers' opinions
(Shahrasbi et al., 2021). Social media activity has been contemplated in
this study following previous research, evidencing it not only as an in­
strument to evaluate the users' perceptions of the waste problem but as a
means to raise awareness and stimulate citizens to take an active part in
the problem-solving process promoting mobile phone waste manage­
ment (Pourabbasi and Shokouhyar, 2022), as well as for promoting
plastic-free behaviors (Heidbreder et al., 2021). However, given that
research usually addresses precise triggers of certain behavior, it be­
comes visible the need to widen the perspective and present a more
general observation of waste management regulations and associated
sentiments, as done in this study. Furthermore, social media have
established to have a significant impact on the creation of people's
recycling behavior (Sujata et al., 2019). As our study confirms, this
importance is equally significant for all categories of the waste sector,
underlining the potential of social media to act as a mediator between
policy-makers and citizens, which can be of substantial importance for
aligning waste management measures and real practices.
Over the last few years, the governments in the Western Balkan na­
tions have made significant efforts in plotting a nation-wide plan to
tackle the growing waste management issue. Among the plans for
building more efficient waste management systems, the North Mace­
donian government also has the intention to further incorporate the
topics of climate change and sustainable development in the overall
national education system, in addition to employing measures to raise
climate awareness among the general public. The ban on single-use
plastic was the starting measure aimed at raising public awareness of
the benefits of waste management regulations. Therefore, it would be
convenient to promote the positive aspect of these measures, initially by
the news media outlets. Since there is a strong positive correlation be­
tween the news reports and the social media response concerning waste
management regulations, more positive news reports informing on the
advantages of waste management, rather than judging individual or
organizational attitudes, may play a substantial role in promoting pro-
environmental propositions, that could be easily exploited by the so­
cial media efficiency for creating public opinion, awareness and action
(Sujata et al., 2019).
5.3. Limitations and future lines of research
The study's limitations encompass data observation restricted to the
case of North Macedonia, which could be overcome in future broader
research, where a comparison could be made for all countries from the
region. Furthermore, a longitudinal study could be helpful to contem­
plate the reaction to different policies implementation and effects
related to climate change generally and not limited to the waste sector
only. Moreover, as previous research has concluded (Lahuerta-Otero
et al., 2018), instead of analyzing only tweets, it can be enriching to
include in the analysis the likes and comments to these tweets, given that
they could, also, provide information regarding the public opinion.
Additionally, current investigation suggests that citizens' socioeconomic
and demographic characteristics might be a relevant predictor of pro-
environmental behavior (Yang and Arhonditsis, 2022), variables
which could be addressed in further research of an identified group of
individuals involved in waste management or other climate-related
causes. Along the same line, the culture of public could play an impor­
tant role in citizens' perceptions, predisposition and engagement in pro-
environmental behaviors, making this element worth of study. Namely,
an identification of Twitter accounts could be attempted, and a survey
could be performed with Twitter users who share and comment on waste
I. Gjorshoska et al. Ecological Informatics 76 (2023) 102130

management-related content so a more complete image of the concerned 20Pollution%20-%20Guidance%20for%20immunization%20waste%20mana

gement_0.pdf
public is generated. Finally, we could employ different methodologies,
Hassen, T., Bilali, H., Allahyari, M., Berjan, S., Radosavac, A., Cvijanovic, D., Vaško, Ž.,
such as, for instance, the PIAR data mining research tool to obtain more 2023. No social distancing from food: how the COVID-19 pandemic shaped student
detailed knowledge about influencers' behavior and communication on food-related activities in the Western Balkans. Nutr. Health. https://doi.org/
Twitter (Lahuerta-Otero and Cordero-Gutiérrez, 2016). 10.1177/02601060221143964.
Heidbreder, L., Lange, M., Reese, G., 2021. #PlasticFreeJuly–analyzing a worldwide
campaign to reduce single-use plastic consumption with twitter. Environ. Commun.
15 (7), 937–953. https://doi.org/10.1080/17524032.2021.1920447.
Declaration of Competing Interest Huang, H., Long, R., Chen, H., Sun, K., Li, Q., 2022. Exploring public attention about
green consumption on Sina Weibo: using text mining and deep learning. Sustain.
Product. Consump. 30, 674–685. https://doi.org/10.1016/j.spc.2021.12.017.
None. Hutto, C., Gilbert, E., 2014. VADER: A Parsimonious Rule-Based Model for Sentiment
Analysis of Social Media Text. https://doi.org/10.1162/COLI_a_00049.
Jenkins, E., Brennan, L., Molenaar, A., McCaffrey, T., 2022. Exploring the application of
Data availability social media in food waste campaigns and interventions: a systematic scoping review
of the academic and grey literature. J. Clean. Prod. 360, 132068 https://doi.org/
Data will be made available on request. 10.1016/J.JCLEPRO.2022.132068.
Jiang, P., Zhou, J., Fan, Y., Klemeš, J., Zheng, M., Varbanov, P., 2021. Data analysis of
resident engagement and sentiments in social media enables better household waste
Acknowledgment segregation and recycling. J. Clean. Prod. 319, 128809 https://doi.org/10.1016/J.
JCLEPRO.2021.128809.
Kittipongvises, S., Phetrak, A., Hongprasith, N., Lohwacharin, J., 2022. Unravelling
This research was supported by the ANSO - Alliance of International capability of municipal wastewater treatment plant in Thailand for microplastics:
Science Organizations Collaborative Research Projects and Training effects of seasonality on detection, fate and transport. J. Environ. Manag. 302,
Projects, grant number ANSO-CR-PP-2022-05. 113990 https://doi.org/10.1016/J.JENVMAN.2021.113990.
Kuai, P., Zhang, X., Zhang, S., Li, J., 2022. Environmental awareness and household
energy saving of Chinese residents: Unity of knowing and doing or easier said than
References done? J. Asian Econ. 82, 101534 https://doi.org/10.1016/J.ASIECO.2022.101534.
Kumar, A., Samadder, S., Kumar, N., Singh, C., 2018. Estimation of the generation rate of
different types of plastic wastes and possible revenue recovery from informal
Adusei, K., Ng, K., Karimi, N., Mahmud, T., Doolittle, E., 2022. Modeling of municipal
recycling. Waste Manag. 79, 781–790. https://doi.org/10.1016/J.
waste disposal behaviors related to meteorological seasons using recurrent neural
WASMAN.2018.08.045.
network LSTM models. Ecol. Inform. 72, 101925 https://doi.org/10.1016/J.
Kurniawan, T., Lo, W., Singh, D., Othman, M., Avtar, R., Hwang, G., Shirazian, S., 2021.
ECOINF.2022.101925.
A societal transition of MSW management in Xiamen (China) toward a circular
Ahamad, N., Ariffin, M., 2018. Assessment of knowledge, attitude and practice towards
economy through integrated waste recycling and technological digitization. Environ.
sustainable consumption among university students in Selangor, Malaysia. Sustain.
Pollut. 277, 116741 https://doi.org/10.1016/J.ENVPOL.2021.116741.
Product. Consump. 16, 88–98. https://doi.org/10.1016/J.SPC.2018.06.006.
Lahath, A., Omar, N., Ali, M., Tseng, M., Yazid, Z., 2021. Exploring food waste during the
Al-Marri, W., Al-Habaibeh, A., Watkins, M., 2018. An investigation into domestic energy
COVID-19 pandemic among Malaysian consumers: the effect of social media,
consumption behaviour and public awareness of renewable energy in Qatar. Sustain.
neuroticism, and impulse buying on food waste. Sustain. Product. Consump. 28,
Cities Soc. 41, 639–646. https://doi.org/10.1016/J.SCS.2018.06.024.
519–531. https://doi.org/10.1016/J.SPC.2021.06.008.
Almulhim, A., 2022. Understanding public awareness and attitudes toward renewable
Lahuerta-Otero, E., Cordero-Gutiérrez, R., 2016. Looking for the perfect tweet. The use of
energy resources in Saudi Arabia. Renew. Energy 192, 572–582. https://doi.org/
data mining techniques to find influencers on twitter. Comput. Hum. Behav. 64,
10.1016/J.RENENE.2022.04.122.
575–583. https://doi.org/10.1016/j.chb.2016.07.035.
Andresen, K., Hoxha, A., Godole, J., 2017. New roles for media in the Western Balkans: a
Lahuerta-Otero, E., Cordero-Gutiérrez, R., De la Prieta-Pintado, F., 2018. Retweet or
study of transitional jounalism. Journal. Stud. 18 (5), 614–628.
like? That is the question. Online Inf. Rev. 42 (5), 562–578. https://doi.org/
Bank, T. W, 2022. Solid Waste Management. Retrieved from. https://www.worldbank.
10.1108/OIR-04-2017-0135.
org/en/topic/urbandevelopment/brief/solid-waste-management.
Licastro, A., Sergi, B.S., 2021. Drivers and barriers to a green economy. A review of
Bird, S., Klein, E., Loper, E., 2009. Natural Language Processing with Python: Analyzing
selected Balkan countries. Clean. Eng. Technol. 4, 100228.
Text with the Natural Language Toolkit. O’Reilly Media, Inc.
Lima, P., Pereira, A., Chaves, G., Meneguelo, A., 2021. Environmental awareness and
Chung, Y., Paik, C., Kim, Y., 2018. Assessment of mitigation pathways of GHG emissions
public perception on carbon capture and storage (CCS) in Brazil. Intern. J.
from the Korean waste sector through 2050. Sustain. Environ. Res. 28 (3), 135–141.
Greenhouse Gas Control 111, 103467. https://doi.org/10.1016/J.
https://doi.org/10.1016/J.SERJ.2017.12.003.
IJGGC.2021.103467.
Directive (EU), 2018. 2018/851 of the European Parliament and of the Council of 30
Little, V., Lee, C., Nair, S., 2019. Macro-demarketing: the key to unlocking unsustainable
May 2018 amending Directive 2008/98/EC on waste.
production and consumption systems? J. Macromark. 39 (2), 166–187. https://doi.
Directive (EU), 2019. 2019/904 of the European Parliament and of the Council of 5 June
org/10.1177/0276146718823885.
2019 on the reduction of the impact of certain plastic products on the environment.
Loureiro, M., Alló, M., 2020. Sensing climate change and energy issues: sentiment and
Dong, X., Lian, Y., 2021. A review of social media-based public opinion analyses:
emotion analysis with social media in the U.K. and Spain. Energy Policy 143,
challenges and recommendations. Technol. Soc. 67 https://doi.org/10.1016/j.
111490. https://doi.org/10.1016/J.ENPOL.2020.111490.
techsoc.2021.101724.
Maiurova, A., Kurniawan, T., Kustikova, M., Bykovskaia, E., Othman, M., Singh, D.,
El Bilali, H.B., 2022. Research on food loss and waste in the Western Balkans: a
Goh, H., 2022. Promoting digital transformation in waste collection service and
systematic review. Front. Nutrit. https://doi.org/10.3389/fnut.2022.983639.
waste recycling in Moscow (Russia): applying a circular economy paradigm to
Elsaid, S., Aghezzaf, E.-H., 2015. A framework for sustainable waste management:
mitigate climate change impacts on the environment. J. Clean. Prod. 354, 131604
challenges and opportunities. (D. Hamid Allaoui Prof, Ed.). Manag. Res. Rev. 38
https://doi.org/10.1016/J.JCLEPRO.2022.131604.
(10), 1086–1097. https://doi.org/10.1108/MRR-11-2014-0264.
Menzel, C., Brom, J., Heidbreder, L., 2021. Explicitly and implicitly measured valence
Ganczewski, G., Jemielniak, D., 2022. Twitter is garbage: a thick big data exploration of
and risk attitudes towards plastic packaging, plastic waste, and microplastic in a
#zerowaste hashtag on twitter in relation to packaging and food packaging
German sample. Sustain. Product. Consump. 28, 1422–1432. https://doi.org/
materials. Packag. Technol. Sci. 35 (12), 893–902. https://doi.org/10.1002/
10.1016/J.SPC.2021.08.016.
pts.2685.
Ministry of Economy, N. M. (n.d.). Strategy for Energy Development of the Republic of
Garcia-Vazquez, E., Garcia-Ael, C., 2021. The invisible enemy. Public knowledge of
North Macedonia up to 2040. Retrieved from https://economy.gov.mk/Upload/Doc
microplastics is needed to face the current microplastics crisis. Sustain. Product.
uments/Adopted%20Energy%20Development%20Strategy_EN.pdf.
Consump. 28, 1076–1089. https://doi.org/10.1016/J.SPC.2021.07.032.
Ministry of Environment and Physical Planning, 2021. План-за-Управување-со-отпад-на-
Georgievski, N., 2021. Single-Use Plastic Bags Banned in North Macedonia,
РСМ-2021-до-2031-година. Retrieved from. shorturl.at/gEOU3.
Biodegradable Ones Sell at €0.25. Meta.mk. Retrieved August 24, 2022, from. htt
Ministry of Environment and Physical Planning, N. M, 2021 November. Long-Term
ps://meta.mk/en/single-use-plastic-bags-banned-in-north-macedonia-biodegrada
Strategy on Climate Action and Action Plan. Retrieved March 15, 2022, from. https
ble-ones-sell-at-e0-25/.
://unfccc.int/sites/default/files/resource/MKD_LTS_Nov2021.pdf.
Gohil, S., Vuik, S., Darzi, A., 2018. Sentiment analysis of health care tweets: review of the
Ministry of Environment and Physical Planning, N. M. (n.d.). Fourth National
methods used. JMIR Public Health Surveill. e5789.
Communication on Climate Change. Retrieved from https://klimatskipromeni.mk
Gollust, S., Seymour, J., Pany, M., Goss, A., Meisel, Z., Grande, D., 2017. Mutual distrust:
/article/32.
perspectives from researchers and policy makers on the research to policy gap in
Nadazdi, A., Naunovic, Z., Ivanisevic, N., 2022. Circular economy in construction and
2013 and recommendations for the future. INQUIRY: J. Health Care Organ. Prov.
demolition waste management in the Western Balkans: a sustainability assessment
Financ. 54 https://doi.org/10.1177/0046958017705465.
framework. Sustainability 14 (2). https://doi.org/10.3390/su14020871.
Grootendorst, M., 2022. BERTopic: Neural Topic Modeling With a Class-Based TF-IDF
Oke, A., Osobajo, O., Obi, L., Omotayo, T., 2020. Rethinking and optimising post-
Procedure. Retrieved from. http://arxiv.org/abs/2203.05794.
consumer packaging waste: a sentiment analysis of consumers’ perceptions towards
Harm, H.C., 2021. Protection Without Pollution >> Guidance for Immunization Waste
the introduction of a deposit refund scheme in Scotland. Waste Manag. 118,
Management. Online: Health Care Without Harm.. Retrieved from. https://nohar
463–470. https://doi.org/10.1016/J.WASMAN.2020.09.008.
m-global.org/sites/default/files/documents-files/6758/Protection%20without%

16
I. Gjorshoska et al.
Online Doc, 2021. Translator. Retrieved August 24, 2022, from. https://www.onlinedo
ctranslator.com/en/.
Paes, M., de Medeiros, G., Mancini, S., Gasol, C., Pons, J., Durany, X., 2020. Transition
towards eco-efficiency in municipal solid waste management to reduce GHG
emissions: the case of Brazil. J. Clean. Prod. 263, 121370 https://doi.org/10.1016/J.
JCLEPRO.2020.121370.
Pedregoa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O.,
Vanderplas, J., 2011. Scikit-learn: machine learning in Python. J. Mach. Learn. Res.
12, 2825–2830.
Pennington, J., Socher, R., Manning, C.D., 2014. GloVe: global vectors for word
representation. In: Proceedings of the 2014 Conference on Emipirical Methods in
Natural Language Processing (EMNLP), pp. 1532–1543.
Perdan, S., Jones, C., Azapagic, A., 2017. Public awareness and acceptance of carbon
capture and utilisation in the UK. Sustain. Product. Consump. 10, 74–84. https://doi.
org/10.1016/J.SPC.2017.01.001.
Polityuk, P., Crellin, F., 2022. Ukraine Reports Higher Chernobyl Radiation after
Russians Capture Plant. Reuters. Retrieved August 24, 2022, from. https://www.reu
ters.com/world/europe/ukraine-nuclear-agency-reports-higher-chernobyl-radiation
-levels-due-heavy-2022-02-25/.
Pourabbasi, M., Shokouhyar, S., 2022. Unveiling a novel model for promoting mobile
phone waste management with a social media data analytical approach. Sustain.
Product. Consump. 29, 546–563. https://doi.org/10.1016/J.SPC.2021.11.003.
Ramachandra, T., Bharath, H., Kulkarni, G., Han, S., 2018. Municipal solid waste:
generation, composition and GHG emissions in Bangalore, India. Renew. Sust. Energ.
Rev. 82, 1122–1136. https://doi.org/10.1016/J.RSER.2017.09.085.
Rayhan, M., Liza, J., Rahman, M., 2022. Assessment of COVID-19 vaccination-related
medical waste management practices in Bangladesh. PLoS One 17 (8 August).
https://doi.org/10.1371/journal.pone.0273053.
Richardson, L., 2007 April. Beautiful Soup documentation. Retrieved August 25, 2022,
from. https://beautiful-soup-4.readthedocs.io/en/latest/.
Roesslein, J., 2020. Tweepy: Twitter for Python!. Retrieved August 24, 2022, from. htt
ps://github.com/tweepy/tweepy.
Shahidzadeh, M., Shokouhyar, S., Javadi, F., Shokoohyar, S., 2022. Unscramble social
media power for waste management: a multilayer deep learning approach. J. Clean.
Prod. 377, 134350 https://doi.org/10.1016/J.JCLEPRO.2022.134350.
Shahrasbi, A., Shokouhyar, S., Zeidyahyaee, N., 2021. Consumers’ behavior towards
electronic wastes from a sustainable development point of view: an exploration of
differences between developed and developing countries. Sustain. Product.
Consump. 28, 1736–1756. https://doi.org/10.1016/j.spc.2021.09.016.
Sidorov, G., Gelbykh, A., Gómez-Adorno, H., Pinto, D., 2014. Soft similarity and soft
cosine measure: similarity of features in vector space model. Comput. Sust. 18 (3),
491–504.
Sujata, M., Khor, K., Ramayah, T., Teoh, A., 2019. The role of social media on recycling
behaviour. Sustain. Product. Consump. 20, 365–374. https://doi.org/10.1016/J.
SPC.2019.08.005.
Taboada, M., Brooke, J., Tofiloski, M., Voll, K., Stede, M., 2011. Lexicon-Based Methods
for Sentiment Analysis. Retrieved from. http://direct.mit.edu/coli/article-pdf/37/2
/267/1798865/coli_a_00049.pdf.
Teh, P., Piao, S., Almansour, M., Ong, H., Ahad, A., 2022. Analysis of popular social
media topics regarding plastic pollution. Sustain. (Switzerl.) 14 (3). https://doi.org/
10.3390/su14031709.
Time.mk, 2022. Retrieved August 25, 2022, from. https://time.mk/.
Time.rs, 2022. Retrieved August 25, 2022, from. https://time.rs/.
Tonnes of COVID-19 Health Care Waste Expose Urgent Need to Improve Waste
Management Systems, 2022. World Health Organization: WHO. Retrieved August
17
Ecological Informatics 76 (2023) 102130
27, 2022, from https://www.who.int/news/item/01-02-2022-tonnes-of-covid-19-
health-care-waste-expose-urgent-need-to-improve-waste-management-systems#:~:
text=Tens%20of%20thousands%20of%20tonnes,to%20a%20new%20WHO%
20report.
Tumasjan, A., Sprenger, T., Sandner, P., Welpe, I., 2010. Predicting elections with
twitter: what 140 characters reveal about political sentiment. Proceed. Intern. AAAI
Conf. Web Soc. Media 4 (10), 178–185.
Vázquez-Rowe, I., Ziegler-Rodriguez, K., Margallo, M., Kahhat, R., Aldaco, R., 2021.
Climate action and food security: strategies to reduce GHG emissions from food loss
and waste in emerging economies. Resour. Conserv. Recycl. 170, 105562 https://
doi.org/10.1016/J.RESCONREC.2021.105562.
Vu, H., Ng, K., Richter, A., Li, J., Hosseinipooya, S., 2022. Impacts of nested forward
validation techniques on machine learning and regression waste disposal time series
models. Ecol. Inform. 72, 101897 https://doi.org/10.1016/J.ECOINF.2022.101897.
Waste Management in Montenegro, 2021 June. Andreas von Schoenberg Consulting.
Retrieved August 20, 2022, from. https://vonschoenberg.info/avscfiles/uplo
ads/2021/06/CMS_Montenegro_Fact_Sheet_final.pdf.
Waste Management in North Macedonia, 2021 June. Andreas von Schoenberg
Consulting. Retrieved August 20, 2022, from. https://vonschoenberg.info/avscfiles
/uploads/2021/06/CMS_North_Macedonia_Fact_Sheet_final.pdf.
Waste Management in the Republic of Serbia, 2021 June. Andreas von Schoenberg
Consulting. Retrieved August 20, 2022, from. https://vonschoenberg.info/avscfiles/
uploads/2021/06/CMS_Serbia_Fact_Sheet_final.pdf.
Waste Statistics - Statistics Explained, 2022. Europa.eu. Retrieved August 20, 2022, from.
https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Waste_statist
ics#Total_waste_generation.
Wei, Y., Gong, P., Zhang, J., Wang, L., 2021. Exploring public opinions on climate change
policy in “big data era”—a case study of the European Union emission trading
system (EU-ETS) based on twitter. Energy Policy 158, 112559. https://doi.org/
10.1016/j.enpol.2021.112559.
Wu, Z., Zhang, Y., Chen, Q., Wang, H., 2021. Attitude of Chinese public towards
municipal solid waste sorting policy: a text mining study. Sci. Total Environ. 756,
142674 https://doi.org/10.1016/J.SCITOTENV.2020.142674.
Wu, Z., Duan, C., Cui, Y., Qin, R., 2023. Consumers’ attitudes toward low-carbon
consumption based on a computational model: evidence from China. Technol.
Forecast. Soc. Chang. 186, 122119 https://doi.org/10.1016/J.
TECHFORE.2022.122119.
Xiao, S., Dong, H., Geng, Y., Fujii, M., Pan, H., 2021. Greenhouse gas emission mitigation
potential from municipal solid waste treatment: a combined SD-LMDI model. Waste
Manag. 120, 725–733. https://doi.org/10.1016/j.wasman.2020.10.040.
Yang, C., Arhonditsis, G., 2022. What are the primary covariates of environmental
attitudes and behaviours in Canada? A national-scale analysis of socioeconomic,
political, and demographic factors. Ecol. Inform. 69, 101661 https://doi.org/
10.1016/J.ECOINF.2022.101661.
Zhang, C., Dong, H., Geng, Y., Song, X., Zhang, T., Zhuang, M., 2022. Carbon neutrality
prediction of municipal solid waste treatment sector under the shared socioeconomic
pathways. Resour. Conserv. Recycl. 186, 106528 https://doi.org/10.1016/J.
RESCONREC.2022.106528.
Zhang, Y., Abbas, M., Iqbal, W., 2021. Analyzing sentiments and attitudes toward carbon
taxation in Europe, USA, South Africa, Canada and Australia. Sustain. Product.
Consump. 28, 241–253. https://doi.org/10.1016/J.SPC.2021.04.010.
Zhou, X., Tao, X., Yong, J., Yang, Z., 2013. Sentiment analysis on tweets for social events.
In: Proceedings of the 2013 IEEE 17th International Conferecence on Computer
Supported Sooperate Work in Design, pp. 557–562.
Од пластични кеси собрани 235 илјади евра, 2022. Retrieved from. https://rb.gy/kdyayy.