Abstract:
This paper shows how the potential for acquiring common sense knowledge from the web can be realized automatically. We describe a system that has been designed to read the web and detect at least some of the kinds of common sense that volunteers have been contributing to Open Mind Common Sense (OMCS). This has at least two significant benefits: by obviating the need to provide certain kinds of knowledge that can be automatically gathered, volunteers can now be directed towards tasks that are less amenable to automation, where their input will be more valuable and effective as a result. Second, it provides a method for broadening the coverage of OMCS knowledge.