Titolo della tesi: How Network Topology Affects the Exposure to Diverse Content: A study to quantify and control the phenomenon
The topology of the Web hyperlinks graph among pages expressing different
opinions may significantly influence the user's exposure to diverse content when visiting more than one page. The evaluation of hyperlink bias is challenging because it depends on the global view network rather than the one-hop neighborhood of individual pages. This thesis introduces algorithmic approaches to quantify and control users' exposure to different aspects of a topic throughout an entire surfing session, rather than just one click ahead. We introduce structural bias, diverse navigability, and exposure to diversity to quantify the phenomenon. On Wikipedia, we evaluate the exposure to diversity on six polarizing topics (e.g., gun control and gun right). We observe that the hyperlink structure often encourages users to remain in a knowledge bubble rather than branch to related aspects of the topic. Then, we devise an insertion algorithm to reduce the network structural bias and a link swapping approach to maximize the diverse navigability. We conclude by presenting a new method to extract the fair densest subgraph to produce unbiased recommendations.