Incremental Parallel Construction Algorithm For The Concept Lattice Of A Real Formal Context

Traditional small-scale data processing methods have been difficult to meet the technical requirements of mass data processing.Therefore,new data processing technologies around big data have followed.As one of the important technologies in big data processing,parallel computing can greatly reduce the program running time and multiply the processing efficiency of big data.Classical formal concept analysis is mainly used in the formal context whose attributes are Boolean-valued.However,in many fields of practical applications,due to the complexity of the problem,attributes are ordinary real-valued in more backgrounds.Although this kind of attributes are more appropriate for describing the uncertainty of practical problem,the high complexity makes traditional algorithms less effective in extracting concepts when the data under consideration is relatively large.With the development and maturity of high-performance parallel technology and parallel computers,it deserves to apply parallel technology in formal concept analysis for improving the efficiency of mining concepts.Based on the real formal context,this paper puts forward an incremental algorithm which is used to extract real-valued concept lattice.And then through analyzing the process of incrementally updating,the algorithm is parallelized.Finally,numerical experiments are conducted to compare serial and parallel algorithms,which verifies the acceleration efficiency of the proposed parallel algorithm.In this paper,the brief research summary can be listed as follows:(1)The method of calculating the real value concept lattice is given,and then a progressive construction algorithm(serial)is proposed for a real value concept lattice calculation.(2)Based on parallel,improved the serial algorithm,and then designed the parallel algorithm.Used MATLAB to implement serial and parallel algorithms.Finally compared the algorithm of numerical experiments between the serial and parallel,parallelization of the algorithm under the real formal condition feasibility were evaluated.