Information sharing and user-generated content on the Internet has given rise to the increased presence of uninteresting and ‘noisy’ information in media streams on many online social networks. Although there is a lot of ‘interesting’ information also shared amongst users, the noise increases the cognitive burden in terms of the users’ abilities to identify what is interesting and may increase the chance of missing content that is useful or important. Additionally, users on such platforms are generally limited to receiving information only from those that they are directly linked to on the social graph, meaning that users exist within distinct content ‘bubbles’, further limiting the chance of receiving interesting and relevant information from outside of the immediate social circle. In this thesis, Twitter is used as a platform for researching methods for deriving “interestingness” through popularity as given by the mechanism of retweeting, which allows information to be propagated further between users on Twitter’s social graph. Retweet behaviours are studied, and features; such as those surrounding Tweet audience, information redundancy, and propagation depth through path-length, are uncovered to help relate retweet action to the underlying social graph and the communities it represents. This culminates in research into a methodology for assigning scores to Tweets based on their ‘quality’, which is validated and shown to perform well in various situations.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:633563 |
| Date | January 2014 |
| Creators | Webberley, William |
| Publisher | Cardiff University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://orca.cf.ac.uk/68758/ |
Page generated in 0.002 seconds