Acquiring Large amounts of Knowledge about Knowledge

In the course of this project we will continue to enhance the capabilities of START from the language perspective. In particular, we will improve its syntactic coverage and extend our existing techniques for parsing and generating a variety of natural languages.

However, the main focus of the proposed work is mechanisms for acquiring large amounts of annotation knowledge, for acquiring ontologies and domain theories and for acquiring expert Know-How, at low cost, with little or no effort. In particular, we will develop methods to:

• Acquire annotation knowledge unobtrusively, as a byproduct of using START, from large numbers of users.
• Acquire annotation knowledge naturally, by adding a knowledge acquistion component to our ``Intelligent Room'' research.
• Acquire annotation knowledge automatically, by automatic discovery of, for example, topic sentences.
• Acquire annotation knowledge directly, by exacting a small ``effort tax'' on those who wish their knowledge to become available via START.
• Acquire ontologies and domain theories by incorporating existing systems using the Protocol of Inference.
• Acquire ontologies and domain theories by allowing users to convey those ontologies and domain theories to START in natural language.
• Acquire expert know-how by interacting with users to debug failed problem solving efforts in which the system gets stuck because of lack of knowledge.

To better illustrate what we have in mind, consider the following scenario. James Smart, our intelligence analyst, uses START constantly to retrieve all sorts of information, including textual intelligence reports, maps, pictures, and the means to communicate with human experts matching descriptions supplied to START. James has also begun to work with HAWK, a system that Helps to Accumulate the World's Knowledge.

• Examples focusing on the kind of knowledge is to be captured
• Examples focusing on the mechanisms that will be used to capture the knowledge
• Examples focusing on the supporting technology that will be deployed to engage the knowledge