Knowledge Discovery Method For Heterogeneous Data Based On Concept Lattice And Its Algorithm Implementation

Nowadays,concept lattice theory as an effective tool of data analysis and processing,has attracted much attention and become a hot research topic.In the concept lattice theory,we mainly studied the knowledge discovery of formal contexts with the same kind of order relations.However,with the arrival of the era of big data,isomerism has become a significant feature of the data.Therefore,it may lose the feasibility and effectiveness when the existing knowledge discovery methods are applied to large-scale data.So,it is urgent to introduce some new knowledge discovery methods to make the concept lattice theory become effective for heterogeneous data sets.Based on the aforementioned discussion,this thesis is mainly to study knowledge discovery methods for heterogeneous data.First,the construction method for heterogeneous concept lattice is given.Second,we discuss the attribute reduct ion problem of heterogeneous formal context subject to preserving the structure of the concept lattice.Moreover,rule acquisition and attribute reduction are investigated in heterogeneous formal decision context as well as an effective algorithm for mining nonredundant rules and reducts.The main contributions of this thesis are as follows:1)The construction method for heterogeneous concept lattice is given.2)The attribute reduction of heterogeneous formal context subject to preser ving the structure of concept lattice is investigated.An algorithm is given,and experiment al results show its effectiveness.3)Rule acquisition and attribute reduction are researched in heterogeneous formal decision context.An algorithm is also designed,and some experiments are conducted to illustrate effectiveness.