Impact of location on social media credibility

Aladhadh, S 2018, Impact of location on social media credibility, Doctor of Philosophy (PhD), Science, RMIT University.

Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Aladhadh.pdf Thesis application/pdf 2.22MB
Title Impact of location on social media credibility
Author(s) Aladhadh, S
Year 2018
Abstract Social media platforms such as Twitter and Facebook allow users from all over the world to contribute content. However, these users publish content without peer review, and contributions of low quality can create credibility concerns. This reduces the potential social benefits of social media. Social media credibility models rely on popularity, temporal patterns and other collective behaviour of users to study the credibility of user generated content (UGC). However, such approaches do not take into account end user credibility perceptions and factors that may influence a contributor (author), which in turn affects credibility models in social media. Therefore, I studied the factors that influence readers' credibility perception and content credibility. I identified a number of limitations in existing models: most research considers only users' perceptions from one country or culture and then generalises the results to others. I also found these models do not consider author location when assessing credibility. Therefore, I proposed a study on the influence of author, reader and event location on user credibility perception and content credibility in social media.

I propose a model that has been validated using a crowdsourced labelling approach. I ran three controlled experiments mainly varying source-based features (author) and content (text). Further, I applied a linguistic analysis approach to validate the influence of location on content credibility. I also applied a number of statistical analyses to measure the effect of all features. I validated the model using a common social media platform (Twitter) and showed the influence of non-textual features on credibility judgments of readers. Also, I found that reader location represented by culture can determine their credibility perception in social media. Moreover, I showed how distance between the event and author location can affect sources and credibility distribution in social media.

Location of readers and authors, and the interaction with event locations can be used to improve assessment of credibility in social media. Reader characteristics are found to be important when studying credibility in social media as they can be used to improve user experience in social media. Moreover, an author's location can enhance credibility detection models to assess content accurately as it can differentiate between content with different credibility levels. While I do not claim that only user location can be used to build a standalone credibility system, I conclude that adding geographic location and culture of users can improve the performance of existing credibility models significantly.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Science
Subjects Information Systems not elsewhere classified
Keyword(s) Data analysis
Statistical analysis
Text mining
Version Filter Type
Access Statistics: 108 Abstract Views, 111 File Downloads  -  Detailed Statistics
Created: Wed, 03 Jul 2019, 14:01:32 EST by Adam Rivett
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us