An integrated framework for detecting online harms, modelling and disrupting of cyberhate networks

Alorainy, Wafa 2022. An integrated framework for detecting online harms, modelling and disrupting of cyberhate networks. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF - Accepted Post-Print Version Download (6MB) | Preview
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (40kB)

Abstract

Hate crimes are not a new phenomenon in society; however, social media and other means of online communication have begun to play an increasing role in hate crimes. Cyberhate, e.g. offensive or antagonistic language targeted at individuals and social groups based on their personal characteristics, which sometimes is considered a form of hate crime, is frequently posted and widely spread via the World Wide Web. The hateful individuals and groups who post this offensive or antagonistic language have increasingly been using the Internet to express their ideas and spread their beliefs. This spread facilitated by the Internet is considered a key risk factor for individual and societal tension leading to regional instability. Automated Web-based cyberhate detection is important for observing and understanding community and regional societal tension - especially in online social networks where posts can be rapidly and widely viewed and disseminated. While previous work has involved using lexicons, bags-of-words, or probabilistic language parsing approaches, they often suffer from a similar issue, which is that cyberhate can be subtle and indirect (or implicit). Thus, depending on the occurrence of individual words or phrases, the analysis can lead to a significant number of false negatives, providing inaccurate representation of the trends in cyberhate. This problem was a motivation to challenge the thinking around the representation of subtle language use, such as references to perceived threats from "the other" including immigration or job prosperity in a hateful context. This thesis does this by proposing a novel "othering" feature set that utilises language use around the concept of "othering" and intergroup threat the- ory to identify these subtleties, implementing a wide range of classification methods using embedding learning to compute semantic distances between parts of speech considered to be part of an "othering" narrative. This novel feature resulted in a noticeable improvement for the classifier performance for both direct and indirect contextual hate. In addition, understanding the network characteristics of these hateful groups could help to understand individuals’ exposure to hate and derive intervention strategies to mitigate the dangers of such networks by disrupting communications. Concerning the people who post hateful content, this thesis analyses their hateful networks in order to build extensive knowledge of hateful group communication. This analysis shows that hateful networks exhibit higher connectivity characteristics when compared to other "risky" networks, which can be seen as a risk in terms of the likelihood of exposure to, and propagation of, cyberhate. This thesis also examines several strategies for disturbing these risky networks. Results show that removing users with a high degree is most effective in reducing the hateful followers’ network connectivity (GC, size and density) and, therefore, reducing the risk of exposure to cyberhate and stemming its propagation. This thesis further reveals that there are notable performance differences between these strategies and their effect on the disruption of hateful networks. The experimental results demonstrated in this thesis contribute to the development of an integrated framework for the countering of cyberhate by proposing a novel feature set for detecting implicit cyberhate, analysing hateful networks and examining several network disrupting strategies.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Computer Science & Informatics
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > T Technology (General)
Date of First Compliant Deposit:	25 August 2022
Date of Acceptance:	19 July 2022
Last Modified:	05 Jan 2024 06:17
URI:	https://orca.cardiff.ac.uk/id/eprint/152153

Actions (repository staff only)

Edit Item