IEEE Technology and Society Magazine - June 2018 - 48

followers who potentially see
a user's tweet. When a user
Tweets
Population (Million)
Participation (Per Million)
tweets, these tweets are added
to the streams of users who folUrdu
Urdu
Persian
low them, so that the tweet may
20
3%
12%
7%
be assumed to be seen and
15
read [2]. Impressions were comPersian
puted per user by adding the
10
17%
results of two expressions: 1)
5
The first was calculated by mulArabic
Arabic
71%
tiplying the number of a user fol90%
Urdu Persian Arabic
lowers by the number of tweets
he/she made; and 2) the second
(a)
(b)
(c)
was determined by the number
of followers who followed the
posts of a specific user and
Figure 1. (a) Pie graph showing proportion of Arabic, Persian, and Urdu tweets. (b)
retweeted by others. Combining
Proportion of real world population of people speaking in different languages. (c) Bar
graph showing participation ratio of Arab, Persian and Urdu society.
topic clustering and impression
factor, we extracted the top ten
influential users in each topic
group. Using different sources, such as their account on
Table 1. User engagement categorized by language.
Twitter and other social networks, such as Facebook,
Instagram, and the social intelligence platform, Klear
Real World
[17], we identified the user type and classified them into
Number of
Population
Participation
four categories: regular users, religious leaders, news
Language
Tweets
(Million)
(Per Million)
agencies/journalists, and special purpose users. The latArabic
5412
365
15
ter refers to accounts created for special purpose, such
as supporting or criticizing religion.
Persian
449
90
5

Urdu

185

Table 2. Topics extracted from tweets.

Topic
Code

Topic
Description

Hijab Advocates

Covering the body is more important than covering
the head

Pictures of incomplete Hijab or Hijab-less women

News

Polling

Hijab does not necessarily mean chastity

Anti-extremism

Hijab Critics

Humors

T10

T11

Porn and misuse of offensive words for advertising

Results
There were 5412 Arabic and 449 Persian tweets regarding hijab, which means that Arabic tweets were 12 times
as frequent as Persian tweets. Assuming the ratio of populations in the real world, the participation of Arab users
is found to be three times as frequent as Persian users.
Persian users tend to retweet news, while Arab users are
more likely to advocate hijab or anti-extremism. Most of
the news published by Persian users concerned a law
voted for by the judiciary-cultural commission of the parliament of the Islamic Republic of Iran, which makes it
illegal for a woman to remove her hijab while driving,
and if the woman is found in violation of the law punishes her by a monetary fine (Figure 3(c)).
The most frequent tweet of Arabic advocates of hijab
was a motto that says: "Liberty neither means dropping
hijab, nor offending religion, nor eliminating morality, it
is promotion of thought, respecting the mind, building a
beautiful future, and a faith to believe and be proud of
it." (Figure 3(a)). Most anti-extremists retweeted a strong
objection to drawing hijab on the picture of a Syrian
refugee mother with her child, who drowned at sea (Figure 3(b) and Figure 4).
We found that fifteen percent (15%) of Arabic tweets
employed our search term used to determine the

IEEE Technology and Society Magazine

∕

JUNE 2018

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2018