What is common sense? If you believe that it is mostly knowledge about the properties of objects, and that it can be acquired by asking questions of strangers, then you will be fascinated by this paper.

Chklovski describes an information collecting system where any Web user can name a topic, T, and then answer intelligent questions about the topic, T. Crucial to the success of this system is that its way of formulating intelligent questions is intelligent. This formulation method is to find the nearest topics to T, and to then ask questions about T for which it knows the answers for most of the neighboring topics. This produces intelligent questions because of the careful choice of the definition of “nearest topic,” and because of the paying of due attention to taxonomic reasoning.

Starting from an original 47147 facts, collected by the Open Mind project, this system now has 217971 facts, from about 3400 contributors. This is still far from the hundreds of millions of facts that may constitute our adult common sense knowledge, but it is enough for a self-organized mapping method to present interesting semantic knowledge of our world conceptualization. I suggest that the methods in this paper can shed some light on this natural question: Does common sense vary among cultures?