Learner: a system for acquiring commonsense knowledge by analogy
Proceedings of the 2nd international conference on Knowledge capture
| Hi-index | 0.00 |
Knowledge is the fuel for intelligent systems. Building software that comes closer to the breadth and flexibility of human reasoning will require substantially larger knowledge bases than any that have been built to date. This talk describes two ways we are exploring for bootstrapping intelligent systems via knowledge capture. First, analogy is useful for knowledge capture because people find it easier to articulate examples than universally valid principles. By exploiting recent advances in cognitive science, we are creating a technology of analogical processing that can (and has) been used with multiple large knowledge bases. Second, sketching is useful for knowledge capture because people find it easier to express many things spatially. By focusing on deeper visual and conceptual understanding of the contents of sketching, instead of recognition, we are building sketching systems that can be broadly applied in many domains. These efforts have benefited substantially from involvement with DARPA research communities, and some lessons learned from those experiences will be discussed.