Detecting Hate Speech within the Terrorist Argument: A Greek Case

摘要

This paper presents a methodology for automatically detecting the presence of hate speech within the terrorist argument. Hate speech can be used by a terrorist group as a means of judging possible targets' guilt and deciding on their punishment, as well as a means of making people to accept acts of terror or even as propaganda for possibly attracting new members. In this paper, we examine both ideology expressed and practices employed by the Revolutionary Organization 17 November (hereafter 17N) that operated in Greece between the years of 1975 and 2002. Within this line of thought, we will focus on the ideological justification, ethical standing and deployment of the terrorist operations as presented in the communiques published by 17N, emphasizing on the use of hate speech as a means of justifying their choices and actions, as well as a way of reaching out to Greek people. To decide on how the automatic classification will be performed, we experimented with different text analyzing techniques such as critical discourse and content analysis and based on the preliminary results of these techniques a classification algorithm is proposed that can classify the communiques in three categories depending on the presence of bate speech. The methodology was tested over the existing dataset with all the communiques and the corresponding results are discussed.