Martina Chichi describes Amnesty International Italy's Barometer of Hate Project, which approaches online hate speech from a human rights perspective. Their goal is to pin down the main targets and triggers for online abuse in Italy, and determine the extent of politician accountability in the level of discourse.
Data Science Salon Vol. 3 on 21 Oct 2019: Social Media – Monitoring Their Impact on Civil Society
Martina Chichi describes Amnesty International Italy's Barometer of Hate Project
1. The Barometer of hatred
and verbal violence in Italy
21/10/2019, Wien
2. 800 segnalazioni 100.000 contenuti valutatirecorded contents recorded contents
6 weeks of
social media
monitoring
3 weeks of
social media
monitoring
600 activists
180 activists,
2.000 hours of
engagement
1.400 candidates'
posts/tweets
461 candidates'
posts/tweets +
comments of
general users
NATIONAL
ELECTIONS 2018
EUROPEAN
ELECTIONS 2019
3. 11,5% (more than 1/10) of contents (posts/tweets + comments/replies)
are offensive and/or discriminatory or hate speech.
Human rights represent a marginal part of the political debate on
social media. The most present topics of this thematic area are migration,
solidarity world (NGOs etc.) and religious minorities.
Politicians' posts and tweets about polarizing topics such as migration and
religious minorities generate a bigger percentage of
offensive/discriminatory/hate speech comments and replies.
4. 0%
37%
10%
32%
21%
50%
40%
48%
40%
33%
The graph doesn't include the LBGTI minority, why?
Occurencies about LGBTI rights/community were too few to have statistical relevance.
The LGBTI minority resulted almost invisible in the political debate on social media and, when
present, the percentage of offensive and/or discriminatory or hate speech comments was high.
Topics with an high percentage of
politicians' offensive and/or
discriminatory posts/tweets collect
more likes, shares and
comments/replies.
Women
Solidarity
Migration
Religious
minorities
Roma
Graph: percentages of offensive and/or
discriminatory or hate speech contents
By politicians By general users
Hate against women is peculiar - even
if politicians doesn't speak negatevely
about this topic, it arouses much
controversy among general users.
Offensive and/or discriminatory
politicians' posts and
tweets generate a bigger percentage
of offensive/discriminatory/hate
speech comments and replies.
5. 0,6%
1,1%
4,6%
0,9%
0,2%
Women
NGOs
Migrantsand
refugess
Muslims
Graph: percentages of
attacks against minorities
Roma
2
74
222
78
43
0,6%
1,1%
4,7%
0,9%
0,2%Focusing only on hate speech (no offensive nor
discriminatory) comments/replies by general
users, these are the 3 main targets:
1. Muslims 2. Migrants and refugees 3. Women
Attacks - general
users (percentage)
Attacks - politicians
(number)
6. Hate can be very explicit in general users' comments and replies,
while it is usually less evident in politicians' posts and tweets.
A typical offensive and/or discriminatory politicians' post/tweet
usually:
1. points the finger at an individual (better if the individual represents
one or more minorities/topics subject of hatred)
2. reports a news about one or more minorities/topics subject
of hatred)
In both cases politicians firstly create a fertile ground for hate
speech and then ask people to share their opinion about the content.
7. Politicians know divisive topics generate more interactions, as well as
negative contents and they adopt consequent strategies. When divisive
topics and/or minorities overlap, hate level can rise even more.
Politicians also invest in hate - hateful posts are a consistent part of
advertisement campaigns on Facebook.
Among 461 monitored politicians 1/2 of the interactions (posts and tweets
+ likes, shares, comments and replies) are generated by one person, Matteo
Salvini. 75% of the interactions is generated by Lega and Fratelli d'Italia,
with an impact on the social media agenda (topics and meanings).