A Quick Algorithm on Generating Concept Lattice for Attribute-Incremental Streaming Data

Yixuan Yang, Fei Hao, Beibei Pang, Keyun Qin, Zheng Pei, Bo Li

Abstract: The explosive information growth facilitates the generation of increasingly massive streaming data. As a traditional soft computing approach for data analysis, Formal Concept Analysis (FCA) has the severe problem of low efficiency and long time for coping with massive streaming data. Therefore, devising an optimized efficiency algorithm of FCA for attribute-incremental streaming data is becoming a very urgent and challenging issue. In this paper, a quick algorithm on generating concept lattice for attribute-incremental streaming data is presented. Specifically, the correctness of the proposed algorithm is mathematically proved, and several real-world data sets are utilized in our experiments in order to compare with the existing two FCA algorithms. Experimental results demonstrate that the proposed algorithm significantly improves the efficiency for generating concept lattice of attribute-incremental streaming data. Particularly, the running time could be reduced by approximately 44% for Jazz dataset.