Frequent Weighted Concept Lattice And Celestial Spectra Associated With The Knowledge Mining System

Concept lattice is an effective tool for the data analysis and knowledge discovery. Weighted concept lattice is a concept lattice structure which introduces the weight value into the intent to depict its importance. Frequent weighted concept lattice is a kind of special weighted concept lattice, whose nodes are all frequent, namely, each satisfies the threshold given by the users. This paper researches on the algebra system of frequent weighted concept lattice, batch constructing algorithm and mining system of weighted association rules.The main research work can be summarized as follows:First, algebra system of frequent weighted concept lattice is constructed. Firstly, frequent weighted concept lattice is proved to be a complete lattice by introducing the concept of virtual node. Secondly, the algebra system of frequent weighted concept lattice is constructed according to the operation of supremum and infimum among frequent weighted concept lattice nodes.Finally, its algebra properties and the complement of knowledge are proved. Second, a batch constructing algorithm of frequent weighted concept lattice is presented. By making use of the most expansibility of weighted concept nodes, frequent nodes, virtual nodes and edges of frequent weighted concept lattice are generated layer by layer from bottom to top, so that the time and storage complexity of constructing the lattice is reduced and the efficiency of batch constructing the frequent weighted concept lattice is improved. In the end, the experiment results validate the correctness and validity of the algorithm by taking the star spectra data and UCI data sets as the formal contexts respectively.Third, on the basis of above, an association rule mining algorithm AWRFWCL based on frequent weighted concept lattice is presented. The mining system of weighted association rules for star spectra data based on the frequent weighted concept lattice are designed and implemented by using VC++ 6.0 and Oracle9i as development tools. The expremental results show that it gives a new approach for pertinency analysis of star spectra data.