I am a fourth-year Computer Science student with an immersion in Mathematics at the Rochester Institute of Technology. I enjoy Fencing, 35mm Film Photography, and Creative Writing. My Research Interests are centered around Data Science and Human Computer Interaction – specifically how to make inferences about the world and objects around around us. I am always excited to lean out of my comfort zone to learn something new, and believe that variation in both perspectives and ideas makes way for tools that can benefit members of a given community. I plan on applying what I learn in my undergraduate degree to an academic career where I can help to build the tools that make technology more accessible and easier to use.
2019 Research Abstract
OPAL: A decentralized solution to data privacy
James Spann1, Dr. Alex ‘Sandy’ Pentland2
1Golisano Collage of Computing and Information Sciences, Rochester Institute of Technology
2Human Dynamics, Department of Media Arts and Sciences, Massachusetts Institute of Technology
Applications on the internet, such as social media or video streaming sites, strive to create an engaging personalized experience but do so at the expense of a user’s personal data. While users can feel as though they have a handle on their data, their personal information can be used in opaque algorithms and distributed without the users’ knowledge. There is no tool that directly keeps track of who is gathering data from a user and how that data is being used, which can lead to applications that take advantage of all of a user’s available data. The focus of this work is to design a protocol system that can minimize the spread of users’ personal data, improve the transparency of the algorithms being used, and allow trust to be represented between users and the platforms that have their data. We use a decentralized approach built on top of modern internet technologies to make an intermediary system that controls who has access to a users data and permission trust amongst other users and their data. This allows users to audit which algorithms were used with their data, while also potentially allowing their data to train federated learning models.