What will a whole nation do? Predicting events in human networks
summary
Research group builds a society-scale graph from Facebook and Twitter plus other sources, encodes each relationship as a long feature vector into an adjacency matrix, trains one classifier per person to do link prediction (binary now, moving to multi-class for hierarchical/effective ties) and runs influence-propagation simulations. Prototype forecasts and monitoring tools exist as proof-of-concept, project Algeria aims for a pilot product in 1–2 years and a later large-scale rollout to augment collective decision making.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:38Terminator
- 9:04Connection of concepts
- 9:52Bob
- 18:42What can we do
- 20:55The main objective
- 21:46Theory
- 24:13Feature vectors
- 25:48Base assumption
- 26:08Influence propagation
- 27:48Link prediction
- 30:46Algeria
related talks
22mFacebook data and machine learning predicting elections?
29mSocial Media Botnet Detection with Linkage Analysis and Machine Learning
13mSocial Physics
41mDemystifying Social Network Analysis (SNA)
43mDehuman future Fifth sun
47mAdvanced Reinforcement Learning & Data Science for Social Impact
24mEmpowering Lifelong Learning: Transforming Data into Insights With GraphRAG
22mGenerating a Billion Personal News Feeds