Exploring the characteristics of abusive behaviour in online social media settings

Alkharashi, Abdulwhab A. (2021) Exploring the characteristics of abusive behaviour in online social media settings. PhD thesis, University of Glasgow.

Full text available as:
[img]
Preview
PDF
Download (8MB) | Preview

Abstract

Online abusive behaviour can impact interaction amongst contributors and moderators. It may lead to physical harm or threats. Existing research has not addressed the perception of moderation activity, discussion and disagreement can cause contributors to react aggressively.

This thesis investigates the factors that lead to abusive behaviour in conversations within online settings. In particular, empirical analyses were conducted to identify the factors that contribute to abuse in online settings and to distinguish between polite and abusive forms of disagreement. Three contributions were presented in this research to address each to social computing, computational social science and cyber abuse research domains.

The analyses suggested that moderators on Reddit view themselves as members of their community and work hard to both guard against violations, but also with contributors to enhance the quality of their content. Moderators also reported the nuances that distinguish polite and abusive disagreement.

Furthermore, the analyses revealed that the differences between in-person and online conversations can help identify abusive behaviour. Specifically, the setting of discussion fosters participant behaviours (less hedging, more extreme sentiment, greater willingness to express personal opinion and straying from topic) that are known to increase the likelihood of abusive behaviour. Additionally, the findings revealed how consensus-building factors can influence disagreement in different settings.

Finally, we showed how disagreement can be identified and can affect votes based on linguistics contexts. It was shown that different forms of disagreement can be detected better when using specific abuse, politeness and sentiment textual features using models of multi label text classification.

The above research findings conceptualised the development of moderation systems to combat online abusive behaviour, based on analysis of the type of disagreement a contribution embodies and other linguistic and behavioural characteristics.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Colleges/Schools: College of Science and Engineering > School of Computing Science
Funder's Name: Engineering and Physical Sciences Research Council (EPSRC)
Supervisor's Name: Storer, Dr. Tim and Jose, Professor Joemon and Hoskins, Professor Andrew
Date of Award: 2021
Depositing User: Theses Team
Unique ID: glathesis:2021-82389
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 17 Aug 2021 09:35
Last Modified: 17 Aug 2021 12:49
Thesis DOI: 10.5525/gla.thesis.82389
URI: http://theses.gla.ac.uk/id/eprint/82389
Related URLs:

Actions (login required)

View Item View Item