Vis enkel innførsel

dc.contributor.authorAttestog, Targeir
dc.contributor.authorKukulage, Samangi Perera
dc.date.accessioned2013-09-24T11:55:18Z
dc.date.available2013-09-24T11:55:18Z
dc.date.issued2013
dc.identifier.urihttp://hdl.handle.net/11250/137569
dc.descriptionMasteroppgave i informasjons- og kommunikasjonsteknologi IKT590 2013 – Universitetet i Agder, Grimstadno_NO
dc.description.abstractPolitical opinions far from what is considered normal, a distorted view of reality, and hatred to certain other groups are spread amongst political extremists like Islamists and White Supremacists. Demonstrations and violence performed by some members of these groups are well-known from mass media and get a lot of attention. The Islamists and right-extremists exploit this benefit to spread their message to ordinary people. In online forums, young, curious people can read detailed information (or propaganda) from extremists. Which words do extremists then use to convince each other in addition to other curious readers that what they stand for is right? The goal of this thesis is to first find algorithms or techniques for how to discover characteristic vocabulary in online extremist forums and words that frequently are used in the same forum message. Then we analyse the results to find patterns of what is typical vocabulary in the different forums. Mapping the extremists’ habits of vocabulary usage can help us know better how extremists write in online extremist forums, and possibly also help us recognize them when they write on some other websites. In this thesis, we find frequent and characteristic words by means of Global Term Frequency (GTF) and pairs of co-occurring words by means of odds ratio in different extremist forums. We compare normalized GTF (NGTF) of words in two forums to find out where they are used most. Words used in only one of two forums are found as well. We find the GTFs for words written by five of the ten most active authors in each forum, and we find words that one author writes, while the other of ten most active authors does not write. From results we see that Islamists write most about religion, but also some politics. Some popular words are “allah”, “prophet”, “fasting”, and “hajj”. The right-extreme websites Stormfront and Vigrid discuss politics and argument for their own ideology and against the mainstream politics. Frequent words in Stormfront are “white”, “jews”, and “race”, in the Norwegian Vigrid website “jødene”, “tyskland”, and “krigen”. In the German right-extreme website Deutsche Stimme, “npd”, “Deutschland”, “partei”, and “volk” are frequent words. Both Islamists and right-extremists are preoccupied by family values. Our results are useful for discovering topics that extremists write about in their online forums, topics that other people do not write about at all or write about with a different point of view.no_NO
dc.language.isoengno_NO
dc.publisherUniversitetet i Agder / University of Agderno_NO
dc.titleMapping extremist forums using text miningno_NO
dc.typeMaster thesisno_NO
dc.source.pagenumberIX, 176 s.no_NO


Tilhørende fil(er)

Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel