Run-time class generation: Algorithms for union of homogeneous and inhomogeneous classes

Abstract

Run-time code (class) generation is a way to develop adaptive intelligent systems, which can dynamically modify their source codes and knowledge bases up to challenges of the working environment. The example of such challenges is a task of integration of new extracted or acquired knowledge into the knowledge base avoiding redundancy of their representation. To execute it effectively the system should be able to analyze and generate (create) new classes of objects dynamically. Therefore algorithms for dynamic creation of new classes of objects via computing the union of homogeneous and inhomogeneous and union of two inhomogeneous classes of objects are proposed in the paper. Proposed algorithms provide an opportunity for knowledge-based systems to generate new classes, which define heterogeneous collections of objects at the run-time; to determine connection between new and previously acquired knowledge; to conclude thematic relevance and connection level of new knowledge with particular theme, category or domain; to integrate new knowledge into the knowledge base. Developed algorithms have quadratic polynomial time complexity and linear space complexity. They can be adapted and integrated into particular object-oriented programming language or knowledge representation model.