A Network, Content, and Sentiment Analysis of Tweets about the Greek

Ministries

IORDANIS KOTZAIVAZOGLOU1, IOANNA PECHLIVANAKI1, DIMITRIOS KYDROS2,

VASILIKI VRANA1

International Hellenic University,

Terma Magnisias, 62124, Serres,

GREECE

Abstract: - Twitter has been widely acknowledged as a crucial platform for political communication and

deliberation. In this sense, research on information extraction from Twitter is growing rapidly but usually uses

sentiment analysis in various aspects. The purpose of this study is to examine the networks of Twitter

interactions among formal and informal political actors, as well as to identify the key topics of discussion and

the sentiments conveyed by users about the 19 Greek ministries, by proposing a combination of three methods

that includes not only sentiment, but also social network and content analysis. The research findings showed

that not all ministries receive equal attention, resulting in interesting differences among them. Such a study can

provide insights into the public’s views, reactions, and concerns, and may help governments and/or ministries

better understand and align their policies and communication with them. In addition, the proposed framework

offers a multifaceted exploration of Twitter interactions, discussions, and sentiments that may be applied

1 Introduction

Twitter, a very well-known and popular social

media platform, has been rebranded as X since late

July 2023 (in the following, we will use the term

Twitter to refer to this platform). Twitter facilitates

political communication and promotes deliberation

between organizations and their stakeholders, [1],

[2]. It has become a crucial tool for participatory

democracy, [3], as it enables formal and informal

political actors to engage in frequent, interactive,

networks, and encourage political expression and

deliberation, [5].

remains a need for further investigation into the

specific context of ministries as public organizations

and their use of social media for political

communication. Limited research has been

conducted on this topic, as evidenced by the scarcity

of studies on ministries in countries such as Greece,

which could provide valuable insights into the field

of social networking analysis and political

communication, [14], [17], [18], [19]. Exploring the

networks of users associated with ministries in

media and political communication in the public

sector. By focusing on Greece as a case study,

researchers can delve into unexplored areas and

generate new insights that can broaden the

understanding of this topic. To our knowledge, no

research has been done to deal not only with Greece

but also with foreign ministries or government

departments.

The paper represents a preliminary effort to

address the research gaps outlined above, with a

Even though there has been some research on

Greek ministries and social media, very little

attention has been paid to the rigorous analysis of

social networking among political actors. Moreover,

the country boasts over 700,000 Twitter users, [20].

In the last couple of years, the Greek ministries have

faced a variety of challenges, including economic

crises, heightened instability, high inflation,

pandemics, tense international relationships, and

upcoming elections, all of which have led to

Twitter data to conclude a vast pool of information.

This approach provides a comprehensive and

created for all 19 Greek ministries?

2 Literature Review

Twitter, due to its interactive, interpersonal, and

dialogic nature alongside its popularity, is

considered a very important medium for

organizations in their communication and

in, shape the public sphere, and express their

opinions about organizations, [27], [28], bringing

about a further democratization of politics, [10].

Twitter allows organizations to interact with users,

and record their preferences and their opinions

following a more direct two-way model of

communication, where organizations as formal

political actors (along with Ministers or politicians)

react, respond, post, like, and share with their

stakeholders in a continuous effort to build relations.

indicated a low level of theme fragmentation, the

introduction of suggestions rather than engaging

voters, and the existence of a strong inverse

relationship between the number of tweets produced

and user interest. In a recent study, [53],

investigated Donald Trump and Jokowi's sentiments

on Twitter regarding Covid-19 policy dissemination.

The outcome demonstrates that both Jokowi and

Trump experienced more unfavorable than positive

sentiments.

sides. The findings demonstrated that there was a

real conflict taking place on social media to affect

people's emotions.

Although sentiment analysis has gained

popularity in recent years, there remains a scarcity

of studies analyzing sentiment in the context of

ministries. For instance, [55], conducted a sentiment

analysis on tweets to assess the public opinion of the

Indonesian Ministry of Health's performance. Their

findings indicated that the vast majority of tweets

develops relationships (edges) in the form of

(re)tweets, mentions (in retweets), and replies.

Networks resulting from Twitter data are directed,

because the content is addressed to specific users.

Understanding the structure of a network is the

point of Social Network Analysis (SNA), [56]. In

the context of the study, finding who communicates

with whom and how their ties develop, could shed

some light on the perceived importance of each

Greek Ministry. The networks created in this step

create networks of words (textual networks)

neighboring each other within texts that are studied

in terms of social network analytic techniques. SNA

focuses not only on the frequency of words in a text

but also deals with the positional properties of the

words to conclude the main topics, the relationship

between words, their proximity, and their

coexistence in a text, [58].

Sentiment analysis uses language processing

techniques to discover meanings, find the sentiment

In both content and sentiment analyses, the

original data were filtered to perform relevant

analyses on tweets, mentions, and replies. Data

include tweets in many languages (Greek, English,

Arabic, etc.), but the analysis focused only on Greek

and partly English, since English ranks highest in

usage on Twitter, [60]. English tweets were

incorporated when they took up at least 20% of the

total comments in a ministry’s network. Wherever

the percentage of tweets was beyond that threshold,

preprocessing was done in raw data, such as stop-

word removal, lemmatizing, etc. After this

with the largest networks on Twitter, ranging from

1423 to 36424 total edges. As one moves down

Table 3, users tend to interact less and less with the

content presented.

Ranking in Table 3 generally also reflects the

interests of the Greek public. It is only natural, for

instance, to find the Ministry of Education and

Religious Affairs on the top spot. Public education

in Greece has undergone frequent changes in recent

years, causing instability for both students and

positions. This problem is particularly acute in the

summer when a surge of traffic is generated on

Twitter as the selection of these teachers takes

place, [62].

Equally expected is the appearance of the

Ministry of Foreign Affairs and Ministry of Defense

in the subsequent places. The geopolitical position

of Greece often brings it face to face with national

or supranational issues pertinent to national defense

and security, energy, etc., [63]. The nation’s defense

countries faced significant backlash from within

their territory, which was also mirrored in online

activity, [64]. Sometimes such issues are so

aggravated that they tread beyond political solutions

and into the military domain. Airspace violations

among countries in the region, for instance, are par

for the course, [65], and sometimes escalate to

armed altercations (e.g., the Kosovo War). All this

justifies the extensive engagement on Twitter

around those two ministries.

COVID-19 outbreak and has since been at the

forefront, largely due to the ongoing health crisis.

Another important observation about networks

concerns the different types of tweets for all 19

Greek ministries’ networks. Types of content are

tweets, retweets, mentions, mentions-in-retweets,

In this phase of the research, an analysis was

conducted to identify the top ten vertices in each

Greek Ministry network based on in- and out-

degree, as well as betweenness centrality.

Subsequently, the accounts were accessed by each

user to discern whether they were formal or

informal political actors. Distinguishing between

personal and business Twitter profiles was a

challenging task, as the platform allows anyone to

create an account and post personal information

consider him a formal political actor, of the

individual type. Ministries, political parties,

municipalities, research centers, institutes,

federations, Universities, museums that belong to

the state, etc., are considered formal political actors,

of the organization type.

Similarly, informal political actors are split into

individuals and organizations. Everyday users that

are not professionally involved in politics are

considered informal political actors, of the

Figure 3 shows the total percentage of formal

and informal political actors, and the media for the

inward-looking edges is not even: formal political

actors rank higher in this metric. On average, 73%

of tweets are targeted toward formal political actors.

We consider such an outcome reasonable because

informal political actors (and media) usually

respond to what formal political actors have to say

and not the other way around. Vertices with high in-

connected nodes. Each ministry’s network is created