Zoning of St. Petersburg Through the Prism of Social Activity Networks

Redrawing St. Petersburg by where people actually gather and post, rather than by its administrative borders.

Landsman D, Kats P, Nenko A, Sobolevsky S · Procedia Computer Science 178: 125–133, 2020 · Read the paper →

A St. Petersburg counterpart to the New York work, asking the same question that runs through all of these: can you recover a city's real structure from the digital traces people leave behind? The traces here are geo-tagged VKontakte posts and Google Places venues, which we used to build networks linking parts of the city that draw the same crowds and the same kinds of places.

Running community detection on those networks carves the city into zones, and the zones line up with St. Petersburg's social and economic geography rather than its official map. It set up the fuller version we published the next year at HICSS. My part was the network analysis and clustering.