Authors:
Atika Mbarek
1
;
Salma Jamoussi
1
;
Anis Charfi
2
and
Abdelmajid Ben Hamadou
1
Affiliations:
1
Multimedia InfoRmation Systems and Advanced Computing Laboratory (MIRACL), University of Sfax, Tunisia, Digital Research Center of Sfax DRCS, 3021, Sfax and Tunisia
;
2
Carnegie Mellon University in Qatar, Doha and Qatar
Keyword(s):
Suicide, Twitter, User Profile, Machine Learning.
Related
Ontology
Subjects/Areas/Topics:
Social Media Analytics
;
Society, e-Business and e-Government
;
Web Information Systems and Technologies
Abstract:
About 800 000 people commit suicide every year and detecting suicidal people remains a challenging issue as mentioned in a number of suicide studies. With the increased use of social media, we witnessed that people talk about their suicide plans or attempts in public on these networks. This paper addresses the problem of suicide prevention by detecting suicidal profiles in social networks and specifically twitter. First, we analyse profiles from twitter and extract various features including account features that are related to the profile and features that are related to the tweets. Second, we introduce our method based on machine learning algorithms to detect suicidal profiles using Twitter data. Then, we use a profile data set consisting of people who have already committed suicide. Experimental results verify the effectiveness of our approachin terms of recall and precision to detect suicidal profiles. Finally, we present a Java based prototype of our work that shows the detectio
n of suicidal profiles.
(More)