Building a healthier feed: Private location trace intersection driven feed recommendations

A paper by Tobin South, Nick Lothian, Taka Yabe, and Alex 'Sandy' Pentland. Presented at the 16th International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction and Behavior Representation in Modeling and Simulation, September 2023.

Structure

This repo has three main sections:

• A python technical demo for matching on location data or images using private set intersection, and how this could be mapped to a feed.
• The figures and reproducible results for the paper.
• A React web demo to show how this could be used in a javascript context. This is being worked to integrate directly with Verdia datastores. Minimal viable code for a website is available here. While separated client and server repositories are available elsewhere (which now use less typescript). These are still being developed and will require you to self host.