PROJECT TITLE :

A Novel Variable Precision Reduction Approach to Comprehensive Knowledge Systems - 2017

ABSTRACT:

A comprehensive information system reveals the intangible insights hidden in an data system by integrating info from multiple information sources in an exceedingly synthetical manner. In this paper, we gift a variable precision reduction theory, underpinned by 2 new ideas: 1) distribution tables and a pair of) genealogical binary trees. Sufficient and necessary conditions to extract comprehensive knowledge from a given data system are also presented and proven. An entire variable precision reduction algorithm is proposed, in that we tend to introduce four important ways, particularly, distribution table abstracting, attribute rank dynamic updating, hierarchical binary classifying, and genealogical tree pruning. The completeness of our algorithm is proven theoretically and its superiority to existing ways for getting complete reducts is demonstrated experimentally. Finally, having getting the whole reduct set, we have a tendency to demonstrate how the relationships between the whole reduct set and the comprehensive information system can be visualized in a very double-layer lattice structure using Hasse diagrams.