Finding Frequent Item Sets from Sparse Matrix
Xiaoyan Zheng, Jizhou Sun · 2009
According to the features of sparse data source while mining association rules, the paper designs a special linked-list unit and two strategies to store data in matrix. A novel algorithm, called SMM (Sparse-Matrix Mining), is proposed to find large item sets from sparse matrix. SMM maps database into a binary sparse matrix and stores compressed data into a linked-list, from which to find large item sets. It uses less I/O and computational time in mining. Experiments show that SMM finds large item sets efficiently and is well scalable.