Pedagogy is an online game for collecting knowledge in a semantically structured form.
You can try our first experimental version of the game on Facebook at http://apps.facebook.com/msr_pedagogy
The web has a lot of interesting knowledge in websites such as Wikipedia. However, most of this knowlegdge is in an unstructured natural language form, which is very easily read by humans, but cannot be meaningfully processed by computers. We would like to stucture this knowledge in a way that allows computers to make inferences and derive useful conclusions from it. An example representation would be an entity-relationship graph, in which entities (people, places, things) are nodes that are connected by edges that represent relations between entities (e.g. "is married to", "was born in",..).
It is a challenging problem to create such semantically structured knowledge bases on a large scale, especially with a rich semantic and syntactic variety of relations. In this project we are exploring an approach based on crowdsourcing with an online game in which players can enter knowledge in a structured form. The player's aim is to "teach" their virtual student about any subjects of their interest, and they get points when the student agrees with other students about the entered knowledge. The incentive for the player is to test how much they know about a subject and how entities are related with each other.
We are currently trying the game out to see what kind of relationships we can extract, and analyse what properties of relations can be inferred.