A new algorithm IUAAR(incremental updating algorithm for association rules)is introduced
by which the change of database records can be classified. The improved FP-tree structure is adopted and the global frequent itemsets are mined by transmitting constrained sub-tree. Utilizing the local FP-tree created by FDMA(fast distributed mining algorithm)only the FP-tree of the added global frequent items is modified. Moreover
using the mined results
the constrained sub-trees of the incremental global frequent itemsets that are transmitted in network are mined. The constrained sub-trees of the original global frequent itemsets can be pruned without transmitting them. Experiments show that in the worst case
IUAAR only scans every local transaction database once
thus the communication cost is dramatically decreased and the maintenance efficiency of global frequent itemsets is improved
and the mining speed of IUAAR algorithm is increased by at least two times in comparison with FDMA.
关键词
Keywords
references
Agrawal R, Srikant R. Fast algorithms for mining association rules [EB/OL]. [2006-05-11]. http:∥www.rsrikant.com/papers/vldb94.pdf.
Cheung D W, Han J W, Ng V T, et al. Maintenance of discovered association rules in large databases: an incremental updating technique [C]∥Proceedings of the 12th International Conference on Data Engineering. Los Alamitos: IEEE Computer Society, 1996: 106-114.
Yi Tong, Xu Baowen, Wu Fangjun. An FP-tree based incremental updating algorithm for mining association rules [J]. Chinese Journal of Computers, 2004,27(5):703-710.
Agrawal R, Shafer J. Parallel mining of association rules [J]. IEEE Trans on Knowledge and Data Engineering, 1996,8(6):962-969.
Schuster A, Wolff R. Communication efficient distributed mining of association rules[C]∥Proceedings of the ACM SIGMOD International Conference on Management of Data. New York: ACM, 2001:473-484.