A deep learning approach for detecting the behavior of people having
personality disorders towards Covid-19 from Twitter
Abstract
This paper proposes an architecture taking advantage of artificial
intelligence and text mining techniques in order to: (i) detect paranoid
people by classifying their set of tweets into two classes
(Paranoid/not-Paranoid), (ii) ensure the surveillance of these people by
classifying their tweets about Covid-19 into two classes (person with
normal behavior, person with inappropriate behavior). These objectives
are achieved using an approach that takes advantage of different
information related to the textual part, user and tweets for features
selection task and deep neural network for the classification task. We
obtained as an F-score rate 70% for the detection of paranoid people
and 73% for the detection of the behavior of these people towards
Covid-19. The obtained results are motivating and encouraging
researchers to improve them given the interest and the importance of
this research axis.