Large Scale Human Sensing Over Time. Challenges and Lessons Learned

Twitter and Facebook continue to be top destinations for information consumption on the Internet. The ever-expanding social graph based enables the implementation of traditional features like item recommendation and selection of trending content that rely on human input and other behavioral data. However, given the enormous amount of human sensing in the world at any given moment in any platform, there is a lot of untapped potential that goes beyond simple applications on top of atomic level content like a post or tweet. In this talk we describe a social knowledge graph that discover relationships as they occur over time and how it can be used to capture the evolution of events or stories.