Please use this identifier to cite or link to this item:
|Characterising the Social Media Temporal Response to External Events
|School of Mathematical Sciences
|In recent years social media has become a crucial component of online information propagation. It is one of the fastest responding mediums to offline events, significantly faster than traditional news services. Popular social media posts can spread rapidly through the internet, potentially spreading misinformation and affecting human beliefs and behaviour. The nature of how social media responds allows inference about events themselves and provides insight into human behavioural characteristics. However, despite its importance, researchers don’t have a strong understanding of the temporal dynamics of this information flow. This thesis aims to improve understanding of the temporal relationship between events, news and associated social media activity. We do this by examining the temporal Twitter response to stimuli for various case studies, primarily based around politics and sporting events. The first part of the thesis focuses on the relationships between Twitter and news media. Using Granger causality, we provide evidence that the social media reaction to events is faster than the traditional news reaction. We also consider how accurately tweet and news volumes can be predicted, given other variables. The second part of the thesis examines information cascades. We show that the decay of retweet rates is well-modelled as a power law with exponential cutoff, providing a better model than the widely used power law. This finding, explained using human prioritisation of tasks, then allows the development of a method to estimate the size of a retweet cascade. The third major part of the thesis concerns tweet clustering methods in response to events. We examine how the likelihood that two tweets are related varies, given the time difference between them, and use this finding to create a clustering method using both textual and temporal information. We also develop a method to estimate the time of the event that caused the corresponding social media reaction.
|Thesis (Ph.D.) -- University of Adelaide, School of Mathematical Sciences, 2019
|This electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at: http://www.adelaide.edu.au/legals
|Appears in Collections:
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.