Knowledge Machine

Background Computers can make reasoned decisions regarding specific domains of expertise through the use of Knowledge Reasoning and Representation (KRR) systems that are populated with domain-specific knowledge bases that include relationships among the knowledge components included in the knowledge base. By allowing the computer to make reasoned decisions on its own, KRR supports automating systems and allows them to make decisions based on a particular set of environmental factors. Once a computer is given a knowledge base, it can develop additional knowledge by connecting pieces of knowledge within the base to form inferences.

A significant complexity of KRR is the need to bring together the domain expertise and the knowledge engineering expertise for a particular domain to encode the knowledge base specific to that domain.

Invention Description Knowledge Machine (KM) and the Component Library solve this problem by creating a library of reusable generic knowledge components, which can then be encoded by a domain expert with minimal training to create a knowledge base for a specific domain. KM is a KRR programming language that allows the user to represent worldly knowledge in a language that the computer understands. The Component Library is a hierarchy of domain-independent knowledge units that can be re-used for each new knowledge base entered.

Unlike other KRR systems, KM and the Component Library do not require a knowledge engineer to implement the system. In addition, the Component Library is easier to program because it uses a smaller set of terms to represent common knowledge than most other KRR systems, yet is able to obtain as broad coverage as other systems through the use of semantics—logical links between different knowledge components.

In a pilot project sponsored by Vulcan Inc. called Project Halo, KM scored a passing grade on an AP Chemistry exam and outscored the other KRR systems in the pilot.

Market Potential/Applications Education, medicine, science, engineering and business applications

Development Stage Commercial product

UT Researcher Bruce Porter, Ph.D., Computer Sciences, The University of Texas at Austin

Type of Offer: Licensing



Next Patent »
« More Software Patents

Share on      


CrowdSell Your Patent