Case Study: NBC News exposes Russian troll networks by analyzing 200,000+ deleted tweets with Neo4j

NBC News Analyzes Hundreds of Thousands of Russian Troll Tweets Using Neo4j

NBC News investigated how Kremlin-linked trolls exploited Twitter during the 2016 U.S. election but faced two big challenges: many troll accounts had been suspended and their tweets deleted, and the sheer volume and anonymity of social media made patterns hard to detect. After the U.S. House committee released a list of suspected accounts, reporters needed to recover missing data and analyze networks of activity to understand how the interference operated.

Using the Neo4j graph database, investigators compiled 202,973 recovered tweets from 454 accounts (sourced from the Wayback Machine and independent monitors) and applied graph algorithms to expose connections, influencers and communities. The analysis showed coordinated networks with a small core of content creators and many retweeters (only ~25% original tweets), clear timing and client-use patterns tied to Moscow business hours, and fake accounts posing as Americans, local media and political groups — findings NBC published in a high-profile exposé.

NBC News

Ben Popken

Senior Business Reporter