Monitoring internet censorship; linguistic connectivity within the webgraph

Thesis

This work offers a significant contribution to the ongoing endeavours in monitoring the effects of internet censorship. It can be freely accessed online by anyone who lives in a censorship free society where limitations on academic texts are not in effect. However, there are numerous places across the globe where this would be highly impractical—due to censorship mechanisms—and potentially illegal. A universal catalogue containing censored pieces of online content, online services and websites does not exist. This thesis discusses new approaches for monitoring internet censorship and the insights gained from experimental analysis with the results.

Key contributions of this work are: firstly, a method for determining if a website is censored in particular country from a remote vantage point; secondly, new approaches for constructing lists of censored domains via a recursive discovery strategy; thirdly, a first look into the relationships between newly discovered censored websites from the perspective of network topology and linguistics.

A number of experiments were conducted to evaluate newly designed frameworks for monitoring censored websites. Using a set of known censored websites from existing lists resulted in the discovery of an order of magnitude more censored material than was previously published. Furthermore, the discovery process yielded useful data and insight into how these censored websites exhibit a multitude of hard and soft connections between them. These results improve the perspicacity of analysis into how online material is censored and give new ways of identifying the motivations and intent of censorship regimes.

These new methods for monitoring internet censorship are significantly more effective than those previously in use, whilst maintaining a strong stance in regard to ethical issues with taking measurements for censorship research.

Authors: Darer, AP

• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: internet censorship; censorship detection; network analysis; censorship; linguistic analysis; webgraph
• Subjects: Censorship; Computer security; Linguistic models; Social sciences--Network analysis