Event Title

Creating Barriers to Determination of Structurally Cohesive Subgroups (Keynote)

Location

King Building 106

Document Type

Presentation

Start Date

4-27-2018 4:30 PM

End Date

4-27-2018 5:20 PM

Abstract

Many activists and human rights defenders have turned to the use of encrypted communication in response to governmental repression and surveillance. However, because encryption use is not predominant, using encryption for activist communication can act as a marker for suspicious activity. Furthermore, corporate surveillance facilitated by private companies is becoming increasingly common, much of which governments could potentially access. This study examines possible methods to prevent the identification of subgroups within a communication network, with a special focus on the security concerns of activists and the broader social context of surveillance. We began by constructing a set of guiding philosophies for methodologies using activist materials and academic reviews. We selected "Fast determination of structurally cohesive subgroups in large networks" by Sinkovits et al. as a model of state-of-the-art subgroup identification abilities, as their proposed algorithm is both exact and significantly faster than prior existing subgroup identification approximation algorithms. Additionally, Sinkovits et al.’s approach has the benefit of a published coded version of the algorithm and benchmark data. A solution algorithm for preventing subgroup identification was then tested on the networks created by said benchmark data.

Keywords:

networks, social networks, privacy of association, subgroup identification, social activism

Notes

KEYNOTE PRESENTATION
Session VI - Technological | Justice

Major

Computer Science; Africana Studies

Award

Mellon Mays Undergraduate Fellowship

Advisor(s)

Robert Geitz, Computer Science
Yveline Alexis, Africana Studies

Project Mentor(s)

Adam Eck, Computer Science

April 2018

This document is currently not available here.

COinS
 
Apr 27th, 4:30 PM Apr 27th, 5:20 PM

Creating Barriers to Determination of Structurally Cohesive Subgroups (Keynote)

King Building 106

Many activists and human rights defenders have turned to the use of encrypted communication in response to governmental repression and surveillance. However, because encryption use is not predominant, using encryption for activist communication can act as a marker for suspicious activity. Furthermore, corporate surveillance facilitated by private companies is becoming increasingly common, much of which governments could potentially access. This study examines possible methods to prevent the identification of subgroups within a communication network, with a special focus on the security concerns of activists and the broader social context of surveillance. We began by constructing a set of guiding philosophies for methodologies using activist materials and academic reviews. We selected "Fast determination of structurally cohesive subgroups in large networks" by Sinkovits et al. as a model of state-of-the-art subgroup identification abilities, as their proposed algorithm is both exact and significantly faster than prior existing subgroup identification approximation algorithms. Additionally, Sinkovits et al.’s approach has the benefit of a published coded version of the algorithm and benchmark data. A solution algorithm for preventing subgroup identification was then tested on the networks created by said benchmark data.