A Framework for Autonomously Performing Knowledge Discovery In Databases
Abstract:
We have tested the hypothesis that the framework used in Lenat's AM discovery program (1982A) may be adapted to provide a framework for autonomously performing knowledge discovery in databases. The proposed framework provides a reasoning component for the autonomous selection of discovery goals, which we believe makes it sufficient for autonomous discovery. In addition, the framework is extremely modular, facilitating the extension of an implementing system. We demonstrated the framework's sufficiency by implementing it in a prototype system called HAMB and using it to make discoveries from the domain of experimental conditions that favor the growth of DNA-protein complex and protein crystals for X-ray crystallographic studies and from the domain of ribosome data-model conflicts. HAMB made many significant discoveries and rediscoveries from both domains. HAMB also demonstrated that its behavior is sensitive to the results of its findings as well as to the user's preferences and background knowledge, necessary conditions for autonomous discovery programs.
